an impact evaluation of india’s second and

An Impact Evaluation of India’s Second and Third Andhra PradeshIrrigation ProjectsA Case of Poverty Reduction with Low Economic Returns

An Impact Evaluation of India’s Second and Third Andhra PradeshIrrigation ProjectsA Case of Poverty Reduction with Low Economic Returns

THE WORLD BANK

45406

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

2006 Annual Report on Operations Evaluation

Annual Review of Development Effectiveness 2006: Getting Results

Addressing the Challenges of Globalization: An Independent Evaluation of the World Bank’s Approach to Global Programs

Assessing World Bank Support for Trade, 1987–2004: An IEG Evaluation

Books, Buildings, and Learning Outcomes: An Impact Evaluation of World Bank Support to Basic Education in Ghana

Brazil: Forging a Strategic Partnership for Results—An OED Evaluation of World Bank Assistance

Bridging Troubled Waters: Assessing the World Bank Water Resources Strategy

Capacity Building in Africa: An OED Evaluation of World Bank Support

China: An Evaluation of World Bank Assistance

The CGIAR at 31: An Independent Meta-Evaluation of the Consultative Group on International Agricultural Research

Committing to Results: Improving the Effectiveness of HIV/AIDS Assistance—An OED Evaluation of the World Bank’sAssistance for HIV/AIDS Control

Country Assistance Evaluation Retrospective: OED Self-Evaluation

Debt Relief for the Poorest: An Evaluation Update of the HIPC Initiative

A Decade of Action in Transport: An Evaluation of World Bank Assistance to the Transport Sector, 1995–2005

The Development Potential of Regional Programs: An Evaluation of World Bank Support of Multicountry Operations

Development Results in Middle-Income Countries: An Evaluation of the World Bank’s Support

Economies in Transition: An OED Evaluation of World Bank Assistance

Engaging with Fragile States: An IEG Review of World Bank Support to Low-Income Countries Under Stress

The Effectiveness of World Bank Support for Community-Based and -Driven Development: An OED Evaluation

Evaluating a Decade of World Bank Gender Policy: 1990–99

Evaluation of World Bank Assistance to Pacific Member Countries, 1992–2002

Extractive Industries and Sustainable Development: An Evaluation of World Bank Group Experience

Financial Sector Assessment Program: IEG Review of the Joint World Bank and IMF Initiative

From Schooling Access to Learning Outcomes: An Unfinished Agenda—An Evaluation of World Bank Support to PrimaryEducation

Hazards of Nature, Risks to Development: An IEG Evaluation of World Bank Assistance for Natural Disasters

How to Build M&E Systems to Support Better Government

IEG Review of World Bank Assistance for Financial Sector Reform

Improving Investment Climates: An Evaluation of World Bank Group Assistance

Improving the Lives of the Poor Through Investment in Cities

Improving the World Bank’s Development Assistance: What Does Evaluation Show?

Maintaining Momentum to 2015: An Impact Evaluation of Interventions to Improve Maternal and Child Health and NutritionOutcomes in Bangladesh

New Renewable Energy: A Review of the World Bank’s Assistance

Pakistan: An Evaluation of the World Bank’s Assistance

Pension Reform and the Development of Pension Systems: An Evaluation of World Bank Assistance

Poland Country Assistance Review: Partnership in a Transition Economy

The Poverty Reduction Strategy Initiative: An Independent Evaluation of the World Bank’s Support Through 2003

The Poverty Reduction Strategy Initiative: Findings from 10 Country Case Studies of World Bank and IMF Support

Power for Development: A Review of the World Bank Group’s Experience with Private Participation in the Electricity Sector

Putting Social Development to Work for the Poor: An OED Review of World Bank Activities

Small States: Making the Most of Development Assistance—A Synthesis of World Bank Findings

Social Funds: Assessing Effectiveness

Sourcebook for Evaluating Global and Regional Partnership Programs

Water Management in Agriculture: Ten Years of World Bank Assistance, 1994–2004

World Bank Assistance to the Financial Sector: A Synthesis of IEG Evaluations

The World Bank in Turkey: 1993–2004—An IEG Country Assistance Evaluation

World Bank Lending for Lines of Credit: An IEG Evaluation

All IEG evaluations are available, in whole or in part, in languages other than English. For our multilingual section, please visit

http://www.worldbank.org/ieg

RECENT IEG PUBLICATIONS

An Impact Evaluation of India’s

Second and Third Andhra Pradesh

Irrigation Projects

A Case of Poverty Reduction with

Low Economic Returns

W O R L D B A N K I N D E P E N D E N T E V A L U A T I O N G R O U P

2008

The World Bank

Washington, D.C.http://www.worldbank.org/ieg

©2008 The International Bank for Reconstruction and Development / The World Bank

1818 H Street NW

Washington DC 20433

Telephone: 202-473-1000

Internet: www.worldbank.org

E-mail: [email protected]

All rights reserved

1 2 3 4 11 10 09 08

This volume is a product of the staff of the International Bank for Reconstruction and Development / The World Bank. The

findings, interpretations, and conclusions expressed in this volume do not necessarily reflect the views of the Executive

Directors of The World Bank or the governments they represent.

The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors,

denominations, and other information shown on any map in this work do not imply any judgement on the part of The

World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries.

Rights and Permissions

The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission

may be a violation of applicable law. The International Bank for Reconstruction and Development / The World Bank

encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly.

For permission to photocopy or reprint any part of this work, please send a request with complete information to the

Copyright Clearance Center Inc., 222 Rosewood Drive, Danvers, MA 01923, USA; telephone: 978-750-8400; fax: 978-750-

4470; Internet: www.copyright.com.

All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher,

The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2422; e-mail: [email protected].

Cover photo: Construction of a main canal in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.

ISBN-13: 978-0-8213-7542-6

e-ISBN-13: 978-0-8213-7543-3

DOI: 10.1596/978-0-8213-7542-6

Library of Congress Cataloging-in-Publication Data have been applied for.

World Bank InfoShop Independent Evaluation Group

E-mail: [email protected] Knowledge Programs and Evaluation Capacity

Telephone: 202-458-5454 Development (IEGKE)

Facsimile: 202-522-1500 E-mail: [email protected]

Telephone: 202-458-4497

Facsimile: 202-522-3125

Printed on Recycled Paper

Contents

i i i

vii Abbreviations

ix Acknowledgments

xi Foreword

xiii Preface

xv Executive Summary

1 1 Introduction3 The Irrigation Paradox6 The Andhra Pradesh Context7 Evaluation Questions7 Study Approach and Overview9 Overview of the Report

11 2 The Projects13 Second Andhra Pradesh Irrigation Project14 Third Andhra Pradesh Irrigation Project 14 Project Implementation16 Summary

19 3 Water User Associations21 Origins of PIM in Andhra Pradesh21 The Current Wave of PIM22 WUA Membership24 What Do WUAs Do and Who Does It?27 Summary

29 4 Economic Benefits31 Direct Production Impacts35 Direct Income Effects: Higher Farm Income36 Indirect Income Effects: Employment Generation38 Who Benefits? Poverty Reduction and Distributional Effects of New Irrigation40 Revisiting the ERR43 Resolving the Irrigation Paradox

45 5 Summary and Lessons Learned47 Findings on the Six Evaluation Questions48 On Selecting Irrigation Investments49 On Participatory Irrigation Management49 On the Impact of Irrigation

51 Appendixes53 A: Sample Selection of IEG Irrigation Survey57 B: Regression Results67 C: Single- and Double-Difference Impact Estimates81 D: Data Tables93 E: The Farm Model, Rate of Return Analysis, and Distributional Impacts

107 F: Village Profiles109 G: Comments from Government of Andhra Pradesh 121 H: Impact Evaluation

123 Endnotes

129 Bibliography

Boxes5 1.1 Irrigation: The Basics7 1.2 Some Local Terminology8 1.3 The IEG Survey

25 3.1 Diversion of Water to Tanks Limits the Number of Beneficiaries43 4.1 Improving the Quality of Sensitivity Analysis in Bank Appraisals

Figures15 2.1 Irrigated Area in SRSP 16 2.2 Irrigation Sources, 2005 and 200622 3.1 Households Accessing Canal and Tank Irrigation Systems Most Likely to

Be in a WUA23 3.2 Likelihood of WUA Membership Increases Sharply with Wealth23 3.3 Large Landowners Most Likely to Be WUA Members; Marginal Farmers

also Active24 3.4 Lorenz Curve for Land Ownership by Irrigation Source26 3.5 Larger Landholders More Likely to Think Water Is Fairly Distributed, as

Are WUA Members32 4.1 Irrigation Increases Yields . . . But by Less than Assumed at Appraisal 33 4.2 Cropping Pattern by Irrigation Source, 2006 35 4.3 Cropping Intensity by Number of Irrigation Sources, 200635 4.4 Increased Cropping Intensity Comes from Additional Growing Seasons38 4.5 Seasonal Employment Effects39 4.6 Distribution of Irrigation Benefits Skewed toward the Better Off40 4.7 Poverty Impact of Irrigation

A N I M PA C T E VA L U AT I O N O F I N D I A’ S S E C O N D A N D T H I R D A N D H R A P R A D E S H I R R I G AT I O N P R O J E C T S

i v

41 4.8 Income Distribution Improved between 2005 and 200643 4.9 Investments in SRBC Had No Chance of Providing an Acceptable Rate of

Return

Tables6 1.1 Bank-Supported Irrigation Projects in Andhra Pradesh

25 3.1 Participation in WUA Activities, 200636 4.1 Crop Failure by Irrigation Source 41 4.2 Estimates of the Economic Rate of Return42 4.3 Alternative Estimates of the Economic Rate of Return

C O N T E N T S

v

Family harvesting crops in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.

v i i

Abbreviations

AP II Second Andhra Pradesh Irrigation Project

AP III Third Andhra Pradesh Irrigation Project

APERP Andhra Pradesh Economic Restructuring Project

APFMIS Andhra Pradesh Farmers Management of Irrigation Systems (Act)

CAD Command area development

CADA Command Area Development Authority

Cusecs Cubic meters a second

DC Distributary committee

ERR Economic rate of return

ha Hectares

I&CADD Irrigation and Command Area Development Department

ICR Implementation Completion Report

IEG Independent Evaluation Group

IMT Irrigation Management Transfer

O&M Operations and maintenance

PIM Participatory irrigation management

Rs Rupees

RLP Rural Livelihoods Project

SAR Staff Appraisal Report

SRBC Srisailam Right Branch Canal

SRSP Sriramasager Project

WOP Without project

WP With project

WUA Water user association

Watering a crop from a field canal in India. Photo by Ray Witlin, courtesy of World Bank Photo Library.

i x

Acknowledgments

The study was made possible by the cooperation

of the government of Andhra Pradesh. The In-

dependent Evaluation Group (IEG) is particu-

larly indebted to Dr. C.V. S. K. Sarma (IAS, Former

Principal Secretary, I&CAD Department, govern-

ment of Andhra Pradesh) and Mr. A. Subba Rao

(Chief Engineer, Sriramasagar Project, I&CAD De-

partment, government of Andhra Pradesh), as

well as many irrigation officials in Warangal. The

staff of the Agricultural Finance Corporation not

only organized the field work but also provided

logistical support to the IEG team in all aspects of

its work. From the Agricultural Finance Corpora-

tion, particular thanks are due to the general man-

ager and project leader for this study, Dr. D. V.

Ramana, to Mr. P. H. Khan (field organizer), Mr. B.

Rajaramakrishna (field supervisor and computer

programmer), and Mr. K. K. Rao (project advisor

and field supervisor); and to field supervisors

Mr. Y. Seshu Ganesh, Mr. C. Manohara, Mr. A.C.

Rao, and Mr. Srinivas Rao. Thanks are also due to

the villagers who took time to reply to question-

naires or to meet with the study team. From the

World Bank, R.S. Pathak has been a useful source

of information and guidance.

This report was prepared by Howard White (task

manager) and Edoardo Masset (consultant), with

inputs from Nina Blöndal (consultant), under the

guidance of Alain Barbu. Useful comments have

been received from Keith Oblitas and Keith Pitman

and the external reviewers Peter Hazell, Michael

Lipton, and Richard Palmer-Jones. The report was

edited by William Hurlbut and Heather Ditt-

brenner. Administrative support was provided by

Soon-Won Pak.

Director-General, Evaluation: Vinod Thomas

Director, Independent Evaluation Group–World Bank: Cheryl Grey

Manager, Sector, Thematic, and Global Evaluation: Alain Barbu

Task Manager: Howard White

Workers digging irrigation ditch in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.

x i

Foreword

Irrigation has made a major contribution to

poverty reduction in the past decades, enabling

higher yields and better nutrition. Despite these

achievements, large-scale irrigation schemes have

usually yielded low returns and attracted negative

publicity because of their adverse environmental

and social impacts. As a result, the Bank has largely

switched its support for irrigation away from new

construction toward rehabilitation and policy re-

form. This evaluation supports the need for reform

but shows that there are substantial benefits from

further investment in infrastructure.

This study analyzes these issues through an im-

pact evaluation of one of the last “old genera-

tion” of projects in which the Bank directly

supported creation of a new irrigation scheme:

India’s Second and Third Andhra Pradesh Irriga-

tion Projects (AP II and AP III). Together these proj-

ects created a new command area, the Srisailem

Right Branch Canal (SRBC), and rehabilitated an

existing one that had been constructed with Bank

assistance, the Sriramasagar Project.

The impact analysis, which uses a quasi-

experimental design and is based on new survey

data and reanalysis of existing data, confirms that

irrigation has a substantial poverty-reducing im-

pact: in the first year of receiving canal irrigation

the poverty rate falls by a quarter. Although larger

land owners benefit the most in absolute terms,

the greatest income growth accrues further down

the income distribution, mainly because of the

growth in wage employment. The long-run

poverty-reducing effects are greater still, as house-

holds become less vulnerable to rainfall fluctua-

tions and so escape the poverty trap of periodic

shocks undermining their asset base.

But the returns to investments anticipated at ap-

praisal did not materialize: Yield increases were

lower than forecast and the expected diversifica-

tion did not occur; indeed, canal-irrigated farms

have been less diversified than those without ir-

rigation. The expectations at appraisal were overly

optimistic on both yield and diversification. The

return was further undermined for AP II and AP

III by long delays and large cost overruns. These

problems were most severe in the case of the

SRBC project, for which the first water was re-

ceived by farmers nearly 20 years after construc-

tion started—resulting in a rate of return of, at

best, zero.

The return to the rehabilitation component was

higher but was still less than 10 percent. In sum-

mary, the investments were efficacious but not ef-

ficient. Investments in large-scale construction

need to establish mechanisms to ensure that con-

struction is completed on time and on budget—

the government of Andhra Pradesh is now

adopting such measures. Part of avoiding delays

is ensuring adequate preparation, which did not

happen in this case, partly because the Bank

raised new design issues once implementation had

begun.

Andhra Pradesh has been at the forefront of par-

ticipatory irrigation management through water

user associations (WUAs). More than 10,000 WUAs

were created across the state, with elections of of-

ficers being held in a single day. These WUAs have

played an important role in improving the qual-

ity of rehabilitation works.

But WUAs cannot take full responsibility for op-

erations and maintenance (O&M); they often lack

the technical capacity to do so. They also lack the

financial resources, as funds collected through the

water cess are insufficient for adequate O&M, es-

pecially as collection rates remain low. In part

this is because the problem of inadequate water

at the tail ends has not been solved. WUAs are not

able to solve problems of water allocation, espe-

cially those between communities. Excess water

use in the head reaches, including sabotage of

water control measures, remains prevalent. There

will continue to be a role for government in water

management, and further means of improving

water allocation need to be considered.

Although the Independent Evaluation Group has

engaged with the government of Andhra Pradesh

throughout the study, the government does not

share all the conclusions of this evaluation. In

particular, it has pointed out that it has taken

steps to rectify the construction inefficiencies

that plagued these projects. The government has

also stressed its continued commitment to de-

veloping WUAs, which it believes will overcome

what it sees as “teething problems” in institu-

tional development.

Among the key findings of the study are the

following:

• Large-scale irrigation can—and in this case did—

have a significant poverty-reduction impact.

• Large-scale construction yields low returns

when there are cost overruns and delays. These

can be avoided with more rationalized selection

and planning and appropriate design and pro-

curement procedures.

• WUAs can help improve water management

but are not a panacea; there is a continued

role for government in both technical aspects

and facilitating equitable water distribution.

x i i


Vinod Thomas

Director-General, Evaluation

x i i i

Preface

The Independent Evaluation Group (IEG) of the

World Bank has undertaken impact evaluations of

the Bank’s support to irrigation in Andhra Pradesh,

India (under AP Irrigation II and III), and of the

U.K. Department for International Development-

supported Rural Livelihoods Project (RLP).

This is one of a series of IEG impact evaluations

(see appendix H). IEG’s program of impact eval-

uation is in part carried out under a Department

for International Development–IEG partnership

agreement; hence the focus on RLP. However,

survey villages are also covered by the Bank-

supported DPIP project, so that the findings are

also relevant to this project.

For this purpose, a first-round survey was con-

ducted in June 2005 in the five RLP districts and

the Warangal district, an area of recent irrigation

expansion along the Kakatiya canal under the Sri-

ramasagar Project, which was rehabilitated with

World Bank support. A second-round survey was

undertaken in Warangal in March 2006, along

with qualitative data collection. This report pres-

ents an analysis of the economic and social impact

of the two irrigation projects.

The report is the first IEG will publish concern-

ing rural development in Andhra Pradesh. Sub-

sequent reports will review the impact of RLP

following a third survey round in June 2007 and

compare the effectiveness of different interven-

tions for reducing rural poverty. This next study

will explicitly compare the effectiveness of large-

scale irrigation investments, the subject of this

study, with the smaller-scale groundwater-based

interventions promoted by RLP, and it will review

the empowerment route to poverty reduction

embodied in both RLP and Indira Kranti Patham.

The findings presented in this report comple-

ment those in the recent IEG study Water Man-

agement in Agriculture: Ten Years of World Bank

Assistance, 1994–2004 (IEG 2006b). Many of the

general lessons in that report concerning prob-

lems encountered in irrigation investments are ex-

emplified by the two projects considered in this

report. Building on the sector review, this report

presents new evidence on the strengths and weak-

nesses of water user associations, which are an im-

portant part of the Bank’s current approach to

irrigation. This study also goes into greater detail

than a typical IEG project assessment on social and

economic impacts, using survey data to identify

the distribution of both direct and indirect ben-

efits from irrigation investments.

A draft of the report was discussed with officials

of the government of Andhra Pradesh in Hyder-

abad in March 2007, and written comments were

subsequently received (attached as appendix G,

with an IEG response to the major points raised).

These comments have been taken into account

as appropriate in the final version of this report.

Building a main brick canal in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.

x v

Executive Summary

For many developing countries, the productivity

of irrigation investments has been falling as a re-

sult of poor operations and maintenance (O&M).

This problem has increasingly been addressed in

development projects by putting users and com-

munities in charge of O&M, an approach known

as participatory irrigation management (PIM).

However, the effect of this approach on equity, and

hence its implications for poverty reduction, have

yet to be fully explored.

The Independent Evaluation Group (IEG) of the

World Bank has examined the paradoxical nature

of irrigation projects by evaluating the impact of

two such projects in the Indian state of Andhra

Pradesh. Irrigation Projects II and III (AP II and AP

III) aimed to irrigate 393,000 hectares (ha) of

land—65,000 ha served by the Srisailam Right

Bank Canal (SRBC), 163,000 ha served by a re-

habilitated canal, and 165,000 ha gained from ex-

tending the system under the Sriramasagar Project

(SRSP). Together these projects were expected to

benefit more than 300,000 farm households.

The evaluation findings have implications for

the management of irrigation projects. Al-

though the World Bank has largely disengaged

from financing new irrigation construction in

India, the findings of the evaluation are of inter-

est, as the government of Andhra Pradesh is plan-

ning major investments in new irrigation schemes.

The findings are also relevant for community-

driven development and other participatory ap-

proaches and to implementation of other reforms

such as water pricing, both of which are part of

the Bank’s current irrigation agenda in India. In-

deed, Bank finance for AP III was originally justi-

fied based on the support it provided to the

reform process in Andhra Pradesh.

Andhra Pradesh has been a leader in the use

of PIM since the late 1990s, when PIM was piloted

in the SRSP. Previously, irrigation O&M was the re-

sponsibility of government agents whose salaries

consumed so much of the O&M budget that lit-

tle went to O&M itself. In 1997, an act of Parlia-

ment handed responsibility for water management

to water user associations (WUAs), which were to

assist in collecting water charges and were to re-

ceive a large share of this revenue to undertake

O&M. The act also provided for the establish-

ment of distributary committees and higher-level

project committees to help manage the O&M.

The World Bank has been closely involved in

these reforms, financing WUA training and irri-

gation system rehabilitation undertaken with the

involvement of WUAs.

Economic ImpactsThe economic impacts of the irrigation proj-

ects were undermined by problems com-

monly encountered in large civil works

programs: cost overruns and construction

The widespread expansion of irrigated farming has largely removed

the threat of malnutrition and premature death for millions around the

world and has had demonstrated poverty-reduction effects. Nonethe-

less, large irrigation projects have fallen out of favor, criticized for their cost-

liness, their environmental impacts, the resettlement problems they cause, and

particularly their poor sustainability record.

delays. AP II did not complete any irrigation

works and thus produced no benefits whatsoever.

AP III completed the investments started under AP

II. Although work started in 1988, no farmers were

receiving water in the SRBC area when AP III

closed in 2004. The planned area will be irrigated

by 2007, nearly 20 years after the project started.

Such long delays make the likelihood of an ac-

ceptable return on investments very remote.

There has been a positive impact on both

crop yields and cropping intensity, but the

expected diversification into high-value

crops has not occurred. An IEG household

survey in a newly irrigated area of the SRSP in the

Warangal district found that irrigated farms are less

diversified than those that are not irrigated. This

lack of diversification is confirmed by other data

sources for different parts of the command area.

This is partly because the water pricing structure

does not discourage excessive use and wastage,

which means that, contrary to official policy, irri-

gated farms grow crops (mostly paddy) requiring

large volumes of water, depriving farms in lower

reaches of water. With the benefit of hindsight, the

expectations at appraisal were overly optimistic

on both yield and diversification.

The bulk of the direct benefits from higher

farm income accrues to the top quarter of

beneficiaries. Indirect benefits from higher em-

ployment are spread more evenly across the in-

come distribution, although the poorest receive

fewest benefits. These indirect benefits in partic-

ular cause irrigation to have a substantial one-off

impact on poverty reduction (on the order of a 10

percentage point reduction in the poverty head-

count as a result of irrigation reaching a village)

and a small progressive impact on distribution. This

is because the greatest income growth from

irrigation is experienced by the second

wealth quartile, a situation that allows a

substantial impact on poverty despite the

skewed distribution of absolute benefits.

The longer-run impact on poverty will be

greater than the immediate impact, because

irrigation ends the cycle of negative income

and debt that undermines asset accumula-

tion. One of the chief benefits of irrigation is the

smoothing effect that it has on incomes, reduc-

ing year-to-year fluctuations in production caused

by variations in rainfall.

The beneficial impacts are less than ex-

pected, as the projects suffered from overly

optimistic appraisals. Yield increases average

about half of appraisal estimates. Hence, given the

other problems encountered by these projects,

the overall return on the investment in SRBC

and SRSP is just under 2 percent (compared with

the appraisal estimate of 24 percent). Even al-

lowing for more optimistic assumptions about

future yields, the economic rate of return does

not rise above 5 percent. Thus, the projects are

a large subsidy to farmers, who do indeed ben-

efit. Moreover, the distribution of benefits is

skewed toward the better off. Herein lies the ex-

planation for the falloff of support for irrigation

despite its poverty-reducing effects: poverty re-

duction is achieved by the large transfer involved

in irrigation financing.

This raises two questions. First, can large-scale ir-

rigation schemes be self-financing and still re-

duce poverty? Second, is subsidizing large-scale

irrigation the best use of a subsidy to reduce rural

poverty? This study answers only the first ques-

tion: schemes can both be self-financing and re-

duce poverty only if construction is efficiently

implemented. Once a scheme is functioning, the

benefits it provides (in good years) are more than

sufficient to pay for the system’s recurrent costs.

These conclusions support the Bank’s shift

to focusing on institutional issues in irri-

gation, including water pricing. The second

question will be the subject of a later IEG study

that considers alternative investments in rural

areas of Andhra Pradesh.

The economic impact findings from this study

are similar to those of other IEG evaluations in the

irrigation sector, including the recent sector re-

view (IEG 2006b): irrigation project outcomes

are restrained by the failure to meet overly

ambitious construction schedules, and in-

x v i


vestment viability relies on overly optimistic

estimates of benefits.

The expected economic rate of return at appraisal

is often far above the ex post rate, which has

rarely been above the 10 percent threshold. The

Bank has responded to these problems in Andhra

Pradesh, as it has elsewhere, by engaging in the

reform agenda, one focus of which has been the

creation of WUAs. Far less progress has been

made on establishing a pricing policy that en-

courages optimal water usage. The emphasis on

WUAs is consistent with the Bank’s emphasis on

community involvement. But a growing body

of evidence, including this report, suggests

that community groups are unlikely to live

up to the hopes invested in them.

Social ImpactsAndhra Pradesh has been a leader in pro-

moting WUAs, an approach that was nur-

tured under AP III. But the effectiveness of

WUAs has not met expectations, and there

is a need for a continuing government role

if the irrigation system is to run efficiently.

The expansion and durability of WUAs has

been influenced by political factors. Fol-

lowing the 1997 act of Parliament, more than

10,000 WUAs were created across Andhra Pradesh.

This occurred in part because the chief minister

at the time had a particular interest in promoting

grassroots structures. However, WUAs ceased to

function when elections were not held in 2002 at

the end of WUA officials’ first term of office. The

planned election of management committees at

the project level in irrigation schemes was shelved

and has still not taken place, although elections

are still planned.

WUA membership is not representative of

the population served. According to the 1997

act, all cultivating households are WUA members.

However, an IEG household survey in 18 villages

in the Warangal district found that half of house-

holds with access to canal or tank irrigation were

members, but just 19 percent of all cultivating

households in the irrigation scheme areas were

members. Membership was strongly skewed to-

ward the better off, and leadership of WUAs was

dominated by members of higher castes (notably

the president, who often runs the WUA with little

consultation). Although supporters of PIM claim

it will benefit the poor, the landless—who are

the poorest—are automatically excluded. Even

among cultivators, it is the better off who are

most likely to participate, partly because they

have better access to irrigated plots.

A positive feature of irrigation development

since the 1997 reforms has been rehabili-

tation of sections of the canal network:

water now reaches more areas. WUAs were

involved in the identification of these works and

undertook most of the rehabilitation. This in-

volvement has improved the effectiveness and

efficiency of these works, even though WUA pres-

idents often did not consult other members and

benefited personally as contractors. The sustain-

ability of the system, however, is open to question.

Even though higher tariffs have been imposed, the

water charges cannot fully finance the amounts

needed for O&M. And with low collection rates,

charges currently fall far short of costs. The re-

forms have not broken the vicious circle whereby

many farmers will not pay for irrigation services

because these services are poor; many others

who do receive water do not pay either.

The farmers’ committee structure (WUAs,

distributary committees, and project com-

mittees) is intended to ensure equitable

allocation of water but has limited capacity

to do this effectively, especially at higher

levels, as the structure is incomplete. The cre-

ation of WUAs and higher-level distributary com-

mittees has not yet been able to prevent the

age-old problem of farmers in the head reaches

of an irrigation system taking more water than in-

tended, thus depriving farmers in the tail reaches.

This problem might have been addressed by proj-

ect committees, but they were never formed, so

the management structure was left incomplete.

None of the tail-reach villages covered by the IEG

study has received irrigation water. Within villages,

canal water is frequently diverted to an existing

E X E C U T I V E S U M M A R Y

x v i i

tank, so farmers drawing water from that tank,

who have their own WUA, are the ones who ben-

efit. Breaches of the canal to take water are also

common, and again WUAs seem powerless to

act. Thus far, most WUAs seem unable to break

the existing pattern of social and political rela-

tionships affecting who can and cannot access

resources. Indeed, WUAs can act as vehicles for

elite capture of these resources.

This report therefore supports the conclu-

sion of IEG’s recent sector study of irriga-

tion (IEG 2006b), which states that the Bank

has been overly optimistic regarding what

may be achieved through WUAs and has un-

derestimated the need for continued gov-

ernment involvement. This finding applies to

technical issues such as the design of rehabilita-

tion works as well as to the ability to intervene in

water disputes and—for the time being at least—

to continue subsidization of system operation.

WUAs are still establishing their roles, so they

may yet fill their potential, but support will be re-

quired for this to succeed.

Lessons Should the Bank avoid financing large irrigation

schemes? This has been the trend, but to abandon

an activity with a large poverty impact is to side-

step the issues involved rather than to resolve

them. For areas where there is unexploited irri-

gation potential, there are still lessons to be learned

about the need to undertake realistic appraisals

to avoid financing projects that are not viable.

On selecting irrigation investmentsMany previous lessons “learned” still remain to be

learned:

• Construction delays undermine the economic

viability of investments, so large-scale con-

struction needs to be accompanied by special

measures (such as those currently being

adopted in Andhra Pradesh) to avoid delays and

cost overruns.

• Unrealistic appraisal estimates increase the

likelihood of poor investments.

• The method of risk assessment used in ap-

praisals understates the actual risk that the

project may not be viable.

• All benefits and costs need to be valued.

On PIM• Support for bottom-up development needs a

strong lead from the top.

• PIM may alleviate the problems of large-scale

irrigation schemes, but it cannot eliminate

them.

• WUAs have only limited means to effectively re-

solve water allocation disputes.

• PIM is not intrinsically pro-poor.

On the impact of irrigation• Irrigation raises output through higher yields

and cropping intensity and stabilizes produc-

tion, all of which contribute to lower poverty.

But irrigation investments constitute a size-

able subsidy to farmers who are not among the

poorest or even the poor.

• The direct income benefits to farmers from

obtaining irrigation provide the potential for full

cost recovery of the recurrent costs of the ser-

vices provided.

• Incentive structures, including water pricing

policy, can encourage water wastage.

x v i i i


Chapter 1Evaluation Highlights• It is paradoxical that irrigation proj-

ects are uneconomical and havenegative social and environmentalconsequence—yet have had strongeffects on poverty reduction.

• Overcoming this paradox requiresaddressing the causes of construc-tion delays and cost overruns andensuring adequate operations andmaintenance.

• Andhra Pradesh did this through leg-islating water user associations, butthe approach faltered and in anycase could not tackle all the prob-lems in the system.

Man walking across irrigated field. Photo by Ray Witlin, courtesy of World Bank Photo Library.

Introduction

This improvement has been made possible by a 30

percent increase in per capita grain production as

irrigated area increased 250 percent. In India, the

expansion of irrigation and adoption of new crops

(the Green Revolution) went hand in hand, with

gross cropped area2 increasing from 23 million to

90 million hectares (World Bank 1997, p. i.).

The Irrigation ParadoxMany studies show the link among irrigation,

higher farm incomes, and substantial multiplier

effects benefiting other rural residents. In India,

poverty rates in irrigated districts are one-third

what they are in districts without irrigation (World

Bank 2005).3 An analysis of poverty trends in 36

villages in Andhra Pradesh found that irrigation

was the main explanatory factor for one-quarter

of all households that had escaped poverty. In

some areas, that figure was more than half

(Krishna and others 2004). Poverty rates are lower,

and social indicators higher, in the coastal re-

gions of the state that have the most irrigation

coverage and have benefited most from major

irrigation schemes.

The Independent Evaluation Group’s (IEG) recent

review of the irrigation subsector identified five

channels through which irrigation can reduce

poverty: direct impact on income, including own

consumption, of poor farmers; increased agri-

cultural employment and wages; multiplier ef-

fects through irrigation-induced growth; lower

food prices; and empowerment through in-

creasing the assets of the poor and their access

to decision making through community organi-

zations (IEG 2006b).

Despite the apparent role that it could play in

achieving the Millennium Development Goals,

irrigation has fallen out of favor. For exam-

ple, World Bank lending for irrigation

peaked in the 1980s and has been

falling since.4 In India, lending fell

from the mid-1980s, and the number

of active projects has remained stag-

nant at around 10 in each year over the

past several decades. Lending rose again after the

mid-1990s, but with a focus on rehabilitation

and institutional reform rather than on new

construction.

This shift in attitude was prompted by, among

other things, inefficiencies within irrigation systems

resulting from poor operations and maintenance

(O&M). Inefficiencies such as cost overruns have

meant that where ex post rates of return have

been calculated, they frequently fall well short of

10 percent (Thakkar 1999).5

In India, for example, IEG rated 8 of 10 irrigation

projects over the years 1991–95 unsatisfactory

on outcome. All 10 suffered implementation

delays and had economic returns well below

the appraisal estimates. The problems included

overly optimistic appraisals with unrealistic time-

frames; poor project design, poor management,

and bad contractors; rent seeking associated with

Irrigation has been a powerful force for poverty reduction and has elimi-

nated the scourge of malnutrition for millions of people. Across the de-

veloping world, average daily calorie availability has risen from 1,930 to 2,6701

over a 42-year period (1961–2003).

3

Though irrigation hasbeen a powerful force forpoverty reduction, it hasfallen from favor.

large-scale public investment programs; and ini-

tiation of follow-up projects before addressing ex-

isting problems.

These problems persisted despite reform attempts

that included the adoption of a more holistic ap-

proach called command area development (CAD).

CAD sought to develop the command area below

the project outlets and organized farmers’ groups

to take part in system management. These

farmer’s groups were the forerunners of the water

user associations (WUAs) that feature in the Bank’s

current support for participatory irrigation man-

agement (PIM), usually with a view to eventual ir-

rigation management transfer (IMT).

Irrigation lending has also been criti-

cized for adverse social and environ-

mental impacts. Some irrigation

projects have been associated with in-

adequate resettlement programs (IEG

1998). Falling groundwater levels and

waterlogging, as well as increasing salinity, have

been among the negative environmental impacts

of irrigation projects, impacts that undermine

the benefits of the schemes themselves. These

problems are acknowledged in the opening para-

graph of the Bank’s 2004 water resources sector

strategy, which states that “50 percent of the

world’s wetlands have disappeared in the past

century . . . with water tables already deep and

dropping every year, and some damaged perma-

nently by salinization” (World Bank 2004a).

It may seem that there is an “irrigation paradox,”

in that an investment that has done so much to

reduce poverty has come to have such a bad

name. And within this paradox is another, which

is that large-scale irrigation schemes (see box

1.1) have attracted the most criticism, even though

they are precisely the schemes supporters say

are needed to bring the benefits of ir-

rigation to all those who need it.

The World Bank attempted to over-

come this paradox by changing its ap-

proach to irrigation. Its new approach

was embodied in its 1993 water resources man-

agement policy, which addressed the negative

environmental impacts and lack of sustainability

that had become apparent. The new strategy

called for an integrated approach to water re-

sources management and for reforms to improve

the relevance, effectiveness, and sustainability of

operations in the main water subsectors.

A 1997 report described these problems in the In-

dian context (World Bank 1997). The emphasis on

new construction had diverted attention from

quality of construction and operations and main-

tenance (O&M). What funds there were for O&M

were inadequate and mostly used for the salaries

of government agents, but government irrigation

departments were inefficient and unaccountable.

Poor-quality irrigation systems set off a vicious

cycle in which cost recovery was weak, as farm-

ers were unwilling to contribute much for poor

services. Further expansion of the system was

constrained physically (many states exhausted

their irrigation potential), environmentally (grow-

ing environmental problems limited productivity),

and fiscally, with irrigation attracting a lower

budget share. The proposed reform program en-

compassed IMT and PIM, privatization of O&M,

raising private funds for new investments, and

appropriate pricing of water and related inputs.

The Bank argued that these far-reaching reforms

could support a second revolution of agricultural

growth through productivity enhancement; the

potential from the resulting scenario was much

greater than from continued area expansion.

IEG’s evaluation of the 1993 strategy found it had

continuing relevance but had been only partially

implemented (IEG 2002). In the irrigation sub-

sector, progress had been made with IMT, but lit-

tle progress was made on cost recovery and

virtually none on addressing other subsidies that

encouraged excessive water consumption,6 such

as low rural electricity tariffs. In line with these find-

ings, the 2004 Water Resources Sector Strategy

(World Bank 2004a) reinforced these messages,

promoting further reform while stressing the im-

portance of water resources for poverty reduction

and emphasizing the need for reforms to en-

courage efficiency in resource use. The strategy

4


Irrigation also haspotentially adverse social

and environmentalimpacts.

Attempts to overcome theparadox have addressed

the reform agenda inwater resourcemanagement.

added that more needed to be done to enhance

the poverty impact of irrigation (World Bank

2004a).

The change in emphasis of Bank lending for ir-

rigation may be the appropriate response to

the “irrigation paradox.” Indeed, there may be

no paradox at all: irrigation achieved great ben-

efits in the past, but now returns are falling as

the scope of major productivity increases has

been exhausted. This would suggest there is

no longer scope for new investments. That is

not entirely true, even in South Asia, and cer-

tainly not in Africa. So consideration of the im-

pact of investing in new infrastructure remains

relevant.

This IEG impact study looks at two Bank-

supported irrigation projects: the Second and

Third Andhra Pradesh Irrigation Projects—AP II

(1988–96) and AP III (1996–2004). The goal of AP

II was to develop a new command area along the

Srisailam Right Bank Canal (SRBC) and to reha-

bilitate and extend the command of the Kakatiya

Canal under the Sriramasagar Project (SRSP).

However, because of various problems, the works

envisaged under AP II were not completed, so AP

III was implemented in the same areas.

I N T R O D U C T I O N

5

Box 1.1: Irrigation: The Basics

Irrigation sources may be either surface water or groundwater.Surface water irrigation draws from natural rivers, artificialcanals, or tanks. Canal irrigation is a treelike system consistingof a main canal and major branches called distributary canals,from which minor (or tertiary) canals spread to reach field canals(or channels).

A canal system is divided into reaches. The head reach is near-est the water source, and the tail reach is farthest from it; the mid-dle reach is between the two. The whole area covered by theirrigation system is called the command area.

Dams are used throughout the system to control flow. Themain canal uses a dam to divert river water from its usual course so it flows into the canal. A gate at the head of eachdistributary and minor canal is used to control the flow of water through the system. Long main canals may have addi-tional dams along their length to control the water flow in thelower reaches. Tanks are “minidams,” collecting water fromvarious sources; that water is then distributed by a minor canalnetwork. Pumping is not usually required using canal or tankwater, as that water flows from the field channels straight intothe fields.

In Andhra Pradesh, irrigation schemes are designated asmajor, medium, and minor and according to their command area.Major schemes are those larger than 10,000 hectares (ha), mediumschemes are between 2,000 and 10,000 ha, and minor schemesare smaller than 2,000 ha.

Groundwater irrigation uses boreholes or tubewells to extractwater from aquifers using pumps. Pumps are now usually elec-tric or diesel operated.

The benefit of large-scale irrigation schemes is that they dis-tribute water over a large area, so that individual areas are lessreliant on local water availability. Rain-fed agriculture that doesnot use irrigation is most sensitive to variations in water supplybecause it depends on rainfall fluctuations. But locally based ir-rigation also depends on the local watershed, and wells may dryup if the groundwater level falls too low.

Irrigation involves complicated resource distribution issues.Capturing water for irrigation deprives users who have been ac-customed to a more plentiful supply. For this reason, the con-struction of major irrigation schemes has become a source ofdispute between states. Yet the presence of a canal or tank raisesthe groundwater level in surrounding areas through seepage. In-deed, in Andhra Pradesh some of the older tanks have been con-verted into percolation dams by blocking the outlets. (A percolationdam or tank is a structure to collect water that then filters into thesoil, recharging the aquifer.)

The costs of irrigation are the cost of constructing the waterstructure, maintenance costs, and charges for water use. In Indiamost of these costs have traditionally been borne by the govern-ment. Revenue is collected from users by the water cess, whichis charged according to the area cultivated rather than the amountof water used.

Amount(US$ ICR/PCR and

Project Dates millions) Command PPAR ratinga

Pochampad Irrigation 1971–79 41 Pochampad Dam and command, now called SRSP n.a.

Godvari Barrages 1976–81 45 Barrage to replace weir for Godvari Estuary n.a.

AP Irrigation and CADComposite Project (AP I) 1976–85 136 Nagarjunasava Reservoir n.a.

AP II 1988–94 118 SRBC and SRSP ICR: Highly unsatisfactory

PPAR: Highlyunsatisfactory

AP III 1997–2004 272 SRBC and SRSP ICR: Satisfactory

Economic Restructuring 1999–2006 142b Support for reform and institutional development Project not yet closedProject across the stateSource: IEG project database.Note: AP = Andhra Pradesh; PCR = Project Completion Report; ICR = Implementation Completion Report; PPAR = Project Performance Assessment Report; n.a. = not applicable. a. The Bank’s ICR, formerly known as the PCR and the PPAR, is an independent review prepared by IEG. Ratings were not assigned in earlier projects. b. Irrigation component only.

The Andhra Pradesh ContextAndhra Pradesh has an extensive irrigation net-

work, including infrastructure constructed with

World Bank support (table 1.1), which suffered

many of the problems outlined above. By the

mid-1990s the state’s irrigation sector was clearly

in crisis. The actual area irrigated had begun to fall

during the 1980s as lack of maintenance left many

distributaries and minor canals unusable. By 1995,

1.54 million ha were irrigated by canals, less than

25 years earlier and more than 15 per-

cent less than the peak reached in 1983.

Under both major and minor schemes

(see box 1.2 for definitions), the state

has an irrigation gap—that is, a gap between the

area for potential irrigation with the existing sys-

tem and the area actually irrigated—of about 1.5

million ha.

The government of Andhra Pradesh produced a

white paper in 1996 identifying a series of prob-

lems, including a decline in net irrigated area as a

result of poor O&M, inefficient and inequitable

water distribution, low yields, and low farm income.

These problems were to be addressed

by the Andhra Pradesh Farmers’ Man-

agement of Irrigation Systems Act

(APFMIS Act), passed by the Legisla-

tive Assembly in April 1997. Water

charges were more than tripled in the same month7

and more than 10,000 WUAs created. The posts in

the WUAs were filled through statewide elections

on a single day in June 1997. Only five months later,

174 distributary committees were created.8 With

the APFMIS Act and the subsequent creation of the

WUAs, Andhra Pradesh was one of the leading

areas in the world implementing PIM.

An early Bank study showcased the experience

(Oblitas and others 1999). The creation of so

many WUAs and the staging of statewide elec-

tions was indeed quite an achievement. How-

ever, nearly half of the elections were uncontested,

at least in part because the government offered

an incentive of 15,000 rupees (Rs 15,000) to each

WUA that unanimously elected all its committee

members and president (Mollinga, Doraiswamy,

and Engbersen 2001).

The philosophy of PIM was consistent with the

Janmabhoomi program launched by the ruling

Telegu Desam Party in 1997. Janmabhoomi was

labeled by the state government as a “people-

centered participatory development process”

through which the chief minister wanted to trans-

form the role of the government from service

provider and regulator to facilitator. Under such

a philosophy, the role of the Irrigation and Com-

6


Irrigation in AndhraPradesh was in crisis by

the mid-1990s.

Sustainability problemsin Andhra Pradesh were

addressed throughlegislation creating WUAs.

Table 1.1: Bank-Supported Irrigation Projects in Andhra Pradesh

mand Area Development Department (I&CADD)

would shift from management of the irrigation sys-

tem to provision of technical support to WUAs.

WUAs would now collect water charges, undertake

O&M, and be responsible for management of the

system—up to command level—through the

three-tier structure of WUAs, distributary com-

mittees, and project committees. However, many

observers saw a political motivation behind the

creation of new structures at the grassroots level;9

they provided an opportunity to build a power

base for the Telegu Desam Party outside of the for-

mal structure of village governments (the Gram

Panchayats), which were dominated by the Con-

gress Party.

However, changing political circumstances tem-

porarily reduced the momentum of PIM. New

WUA elections were not held on schedule when

the terms of office of the WUA officers elected in

1997 expired; many WUAs became moribund

until elections eventually took place in 2005–06.10

This momentum has been somewhat restored

in the last two years. However, the new Congress

Party government came to power on a platform

of promoting agricultural development, central to

which are new irrigation schemes. Unlike many

other areas in India, Andhra Pradesh has reached

only about half its irrigation potential. The latest

five-year plan envisages substantial investments in

new irrigation (appendix table D.1). But the Bank

has been reluctant to commit to a substantial

program without improved efficiency in the ex-

isting system. This evaluation thus feeds directly

into the very relevant question of the viability of

large-scale irrigation investments in Andhra

Pradesh. Central to the study, therefore, are the

underlying questions on the role of the Bank in

promoting PIM and how the adoption of PIM im-

proves the viability of irrigation investments.

Evaluation QuestionsThe study considers six questions:

• Do investments in irrigated agriculture increase

production in Andhra Pradesh?

• What are the impacts of PIM on incomes and

poverty alleviation, and who benefits?

• Is investment in major canal irrigation

economical?

• Was the Bank’s support for canal irrigation rel-

evant and appropriate?

• Do WUAs facilitate greater participation by the

poor and by disadvantaged groups in water

management and access to irrigation?

• Are WUAs an effective means of providing sus-

tainable O&M?

Study Approach and OverviewThe study uses a quasi-experimental design, as did

IEG’s recent impact evaluations (IEG 2006b, 2005,

2004, 2000). Undertaking an impact study for a

project that had closed and for which there were

no baseline data was a challenge.11 But although

the SRBC project closed in 2004, many farmers

were still not connected to water in

the new command area. The IEG team

was informed that approximately

44,000 ha of the SRBC command would

be irrigated by the 2005 kharif season,

with another 21,000 ha to be connected


7

Box 1.2: Some Local Terminology

Caste: Castes are hereditary social groups, the members of which weretraditionally restricted in the activities they could perform. The 1950 con-stitution of a newly independent India included a schedule of castes andtribes thought to be the most disadvantaged—hence the name scheduledtribes and castes—who benefited from reservation of certain positions,including members of Parliament, where they alone were allowed to stand.The 1980 Mandal Commission proposed to extend reservation to other back-ward castes, that is, group less well off than the remaining other castes,though not as underprivileged as the scheduled tribes and castes. Theserecommendations were implemented, against considerable protests, in 1990.

Mandals: Each state is divided into districts and each district further di-vided into subdistricts, which in Andhra Pradesh are called mandals.Andhra Pradesh has 18 districts and 786 mandals.

Rain and Cropping Season: There are two main cropping seasons inAndhra Pradesh: kharif, for which planting is immediately after the mon-soon in July/August, and rabi, in the first months of the year. Rain-fed areashave only a kharif season, whereas those with irrigation may plant inrabi and possibly a third, summer, season. Alternatively, some crops canbe cultivated twice a year but do not fall within a specific season.

Andhra Pradesh still hasirrigation potential, butthe Bank has drawnattention to the need toaddress efficiency.

the following year. This phasing-in of irrigation

suited IEG’s sample design, which was based on

the “pipeline approach” of selecting control areas

to be treated that were as yet untreated (see

appendix A).12

The first round of the survey, carried out

in June 2005 (see box 1.3), would in ef-

fect be a baseline for areas to be irri-

gated (treated) for the first time in the

month immediately after the survey.

The second round of the survey, in

March 2006, would capture the impact of the first

two cropping seasons (kharif and rabi) for these

newly irrigated areas. Unfortunately for both the

study and the farmers, the Irrigation Department’s

figures were something of an exaggeration (see

chapter 2). A far smaller area of the SRBC command

had been irrigated—and this by test waters—and

the full command was only due to be completed

by mid-2007. The lack of irrigation under the SRBC

command made it unsuitable for the survey; in-

deed, it means that impact estimates for the SRBC

command cannot be based on actual data.

The survey was carried out in the Warangal dis-

trict, toward the tail reaches of the SRSP command

below Lower Manair Dam.13 This system had been

both rehabilitated and extended under AP II and

AP III. But five years of drought meant that in the

2004–05 season no water was released in the

SRSP canal for the first time since it had opened.

Groundwater levels were low so that other irri-

gation sources were also yielding less than in pre-

vious years.

In contrast, rainfall from the monsoon season of

2005–06 was very good. The survey provides

ample evidence of how the rains improved agri-

cultural performance and reduced poverty. But

these factors do make it more of a challenge to

identify the impact of the irrigation projects. This

difficulty is compounded by the lack of mea-

surement of water releases below the head of

the distributary canal (and even these data are not

compiled centrally), so there are no accurate

means of measuring water use. For evaluators, the

disruption of canal water supplies in 2004–05

provides a “natural experiment,” 14 whereby a

before-versus-after comparison (in fact, after ver-

sus before) of normally irrigated areas can provide

additional indicators of irrigation impact.

The use of data from two survey rounds allows for

double-difference estimates of project impact to

be made, that is, impact calculated as the change

in outcome in the project area less the change in

outcome in the comparison group (see appendix

C). Many households received canal water through

recharging of the village tanks. Hence, project

beneficiaries are identified as those using both

canal and tank waters.

Comparisons between beneficiaries and non-

beneficiaries were made at three levels: plot,

household, and village. For the plot and house-

hold comparisons, comparisons were made be-

tween those who had newly acquired canal or tank

irrigation between the two rounds of the survey

and those who had not. Those who had not made

up two separate comparison groups: those who

had no irrigation at all in 2005 or 2006 and those

who continued to use the same source. The for-

mer is the more appropriate control and is best

compared with the subsample of the treatment

group that had no irrigation in 2005. However, the

sample size does not always permit this compar-

ison. The comparisons of village-level averages

compare villages that received canal irrigation for

8


Box 1.3: The IEG Survey

The IEG survey consisted of two rounds: one in June 2005 and the sec-ond in March 2006. The first round, just before the 2005 kharif season started,asked questions about production in three preceding seasons (2004 kharifand 2004–05 rabi and summer 2005). The second survey asked about thethree subsequent seasons: kharif in 2005, rabi in 2005–06, and expecta-tions regarding summer 2006. Fifty households were surveyed in 18 villages,making a total sample size of 900 households. The same households wereinterviewed in both rounds. A village questionnaire and a questionnairefor the WUA chairman were included. A separate qualitative study teamwas fielded in six villages in March 2006.

Data were collected on income, farm budgets, irrigation, and variousaspects of social and political life in the village.

The analysis presented in this report uses data from a number ofsources—both primary and secondary—in addition to the IEG survey.

The Andhra Pradeshprojects offered an

opportunity for a before-and-after impact

analysis.

the first time between the two surveys with villages

that have no access to canal irrigation.15

The sample for IEG’s survey was selected to fa-

cilitate examination of the impact of new irri-

gation. But the surveyed area is clearly not

representative of the project area as a whole; the

area below Lower Manair Dam is notably different

in cropping patterns because of a relative contin-

ued water scarcity. Therefore, the impact analysis

in this report is based on a number of different

sources, such as the AP III baseline data, various

data from the government of Andhra Pradesh,

and the figures used by World Bank operational

staff in the completion report for AP III.

Overview of the ReportChapter 2 describes the two projects studied: AP

II and AP III. The following two chapters discuss

the workings of WUAs (chapter 3) and the eco-

nomic impact of the project (chapter 4). Chapter

5 lists lessons learned.


9

Chapter 2Evaluation Highlights• AP II suffered from a faulty and in-

complete appraisal and failed.• AP III aimed to complete the work

started by AP II.• AP III adopted a participatory pro-

cess that relied on WUAs for irriga-tion management.

• The works supported by the projectexpanded the area under irrigation,but many farmers have yet to receivewater, nearly 20 years after construc-tion started.

• Despite improved service, water al-location issues still exist; for example,irrigation water often does not reachtail-enders.

Indian women working in cotton field. Photo by Ray Witlin, courtesy of World Bank Photo Library.

The Projects

Second Andhra Pradesh Irrigation ProjectFirst, SRSP was a renovation and extension of the

system, with modernization of 165,000 hectares

(ha) in the command area already supported by

the Bank under the Pochampad Irrigation Project

(1971– 79). A further 163,000 ha were to be de-

veloped below the Lower Manair Dam, located at

kilometer (km) 146 on the Katakiya canal. Second,

the SBRC irrigation scheme involved construct-

ing a main canal 143 km long with a capacity of

750 cubic feet per second, with block develop-

ment to irrigate an area of about 65,000 ha.

These main canal works were to be supplemented

by supporting activities, such as the provision of

training facilities and training of professional and

support staff required for project implementa-

tion, training of farmers, resettlement activities,

and construction of feeder roads.

In accordance with the project objectives, the

principal project benefits expected were higher

agricultural value and farm employment. However,

these benefits were not realized because the proj-

ect ran into serious construction problems. There

were both substantial cost overruns (of 326 per-

cent for SRSP and 175 percent for SRBC) and de-

lays in carrying out work. Although some major

construction was implemented, or at least partly

implemented, no contiguous portion of new or

rehabilitated canal or command area was com-

pleted before the project closed. Just over one-

third (38 percent) of the planned civil works were

completed in the SRBC command, and only about

one-fifth (21 percent) of those planned in SRSP

were completed. Thus, no irrigation—either mod-

ernized or new—was provided under AP II.

The project’s failure can be attributed

to problems in both design and im-

plementation. The design underesti-

mated costs, notably for the amount of

hard rock excavation that was required to con-

struct the SRBC main canal. The problem was

made worse by government-estimated unit costs

(used in the project cost estimates) that were

well below the market values for such works. The

selection of contractors was poorly done, and

they performed badly, exacerbating the cost over-

runs. Project documents also speak of “rent seek-

ing” associated with construction activities,

although no direct evidence is provided.1

Project design also took inadequate account of two

issues. The first was the difficulty of acquiring

land. Second, because water supply was limited,

the state government’s policy was that all new sur-

face irrigation projects in the two regions would

be designed and operated to satisfy the require-

ments of irrigated dry crops, requiring an en-

forced change in the cropping pattern because of

reduced water allocation (a policy known as lo-

calization, which dated from colonial times).

Specifically, under AP II, water would be required

to fill Lower Manair Dam to provide irrigation in

the newly developed areas below the dam. This

water previously had been available to farmers in

AP II started in 1988 and was the fourth Bank-supported irrigation proj-

ect in Andhra Pradesh (see table 1.1). With the objective of increas-

ing agricultural production and farm incomes, AP II was to bring

irrigation to two poor, drought-prone areas through two subcomponents.

1 3

AP II ran into severeconstruction problemsand failed.

the SRSP command above Lower Manair Dam, so

those farmers were accustomed to irrigated wet

agriculture, notably paddy, which is the preferred

crop in the region.

However, attempts to force changes

in cropping patterns have a poor record

in Andhra Pradesh, including the fail-

ure of this policy in an area adjacent to

SRBC in the first irrigation scheme in

Andhra Pradesh. These difficulties were

a manifestation of a problem well en-

trenched in Indian irrigation. Schemes

are designed with the widest possible coverage,

partly to help ensure food security and partly for

political reasons. But the system design is overly

optimistic, not taking account of likely cropping

patterns and actual system losses. Hence, there

is insufficient water for all users. In the absence

of allocation through pricing, the result is head-

enders taking more water and tail-enders going

without.2 There has been conflict at times.

Overall, the Bank’s preproject assessment of AP

II was severely flawed: the Bank accepted the

government’s inaccurate cost estimates and a de-

sign the Bank itself later questioned. Moreover,

the Bank also accepted projections of the area to

be covered, future yields, and crop mix, all of

which proved optimistic—a situation that should

have been expected given earlier experience.3

Substantial problems were also caused by the

Bank’s management of the project. Design

changes were required after the first contracts had

been issued. It took time to come up with the re-

vised designs, but the Bank then commissioned

studies that supported the original design. There

was a protracted dispute over the sufficiency of

the water allocation for SRBC, during

which the Bank informally suspended

the credit. Finally, lack of familiarity

with Bank procedures among govern-

ment officials slowed procurement.

With slow progress, the Bank cancelled $145 mil-

lion of its $271 million contribution to the project.

However, there was agreement that a new project

(AP III) would support completion of the works.

Third Andhra Pradesh Irrigation ProjectThe overall objective of AP III (1996–2004) was

to complete the irrigation development and re-

habilitation work begun under the previous proj-

ect and thus realize the potential for increasing

agricultural productivity and rural incomes in

two economically backward regions of Andhra

Pradesh. This was to be achieved through a num-

ber of activities.

The project was scaled back, with the following

elements dropped: irrigation development and re-

habilitation works (80 percent of cost) for com-

pletion of the irrigation network and feeder roads

of the 65,000-ha SRBC irrigation project; rehabil-

itation of 253,000 ha of the SRSP scheme’s irri-

gation system, of which 88,000 ha were below

Lower Manair Dam, up to km 234; and extension

of the canal beyond km 234.

Agricultural support services (2.5 percent of cost)

would foster agricultural diversification and pro-

ductivity through irrigation agronomy research

and improved command area extension services,

and improve irrigation services by PIM through

the establishment of WUAs. This component

had three subcomponents: (i) an irrigation agron-

omy program, (ii) a WUA promotion program, and

(iii) a farmer training program. Support to WUAs

in the SRSP command was, for the Bank, piloting

an approach that was to be expanded statewide

in the irrigation component of the Andhra Pradesh

Economic Restructuring Project.4 Activities sup-

ported under AP III were expected to serve as a

model for the rest of the state with respect to as-

pects such as collection of water charges.

The project also included a resettlement and re-

habilitation program (5 percent of cost), an en-

vironmental management plan (7.5 percent of

cost), and dam rehabilitation and safety assurance

works (5 percent of cost).

Project Implementation

SRSPThe rehabilitation works for SRSP were com-

pleted. This resulted in an increase in the carry-

ing capacity of the main canal and the command

1 4


Costs wereunderestimated and

the difficulty ofacquiring land and

of achieving a changein cropping patterns

was neglected.

AP III was undertakento complete the

unfinished work of AP II.

irrigated area of 115,000 ha from 1995–99 (fig-

ure 2.1); by 2005, the area irrigated above Lower

Manair Dam had reached the target of 165,000 ha.

Rehabilitation works above Lower Manair Dam

have enabled water to flow for a greater distance

down distributaries and minor canals, so that

more areas get water. Actual irrigation in the SRSP

command as a percentage of that planned rose

from an average of 50 percent in 1995–96 to

nearly 90 percent five years later (appendix table

D.2). The works below Lower Manair Dam have

been completed, with the most recent data show-

ing an irrigated area of 90,000 ha, slightly above

the 88,000 ha expected at appraisal.5

However, it is likely that the additional irrigated

area is in fact short of that expected under AP III.

The figures reported here are for the area I&CADD

believes will be irrigated, which overestimates

the actual area irrigated. Revenue Department

estimates of irrigated area are typically 40 percent

lower than those of I&CADD (Jairath 2000).6 Al-

though Revenue Department estimates may be bi-

ased downward,7 I&CADD figures are likely to

overestimate the area by at least 20 percent. The

actual area covered is invariably less than the po-

tential reported, owing to the failure of water to

reach tail-enders because of system losses (a well-

known problem, confirmed by the field work for

this study) and excess use by farmers in higher

reaches.8

Excess use in higher reaches is discussed later in

this report. Regarding system losses, the “baseline”

survey found the water flow in the main canal to

be 90 percent of capacity on average, ranging

from 100 percent in higher reaches to 76 per-

cent toward the tail end. The average capacity of

the distributaries is less. The failure to realize full

capacity results from buildup of silt, land slip-

pages around cuttings and embankments, and

damaged linings. These data were collected shortly

after the completion of the rehabilitation financed

by AP III, so the situation is likely to have wors-

ened since then.

Nonetheless, the greater carrying capacity of the

main canal has made possible the expansion of the

command area; under SRSP Stage II,

the state government is extending the

main canal a further 62 km, the dis-

tributaries of which will eventually ir-

rigate 440,000 ha. In the areas covered

by IEG’s survey, which included the newly irrigated

areas between km 234 and km 284 of the main

canal, the area irrigated by canal or tank doubled

from 13 percent of all cultivated land in 2004–05

to 26 percent in 2005–06 (figure 2.2; see also ap-

pendix table D.3).

The bulk of this increase is from tanks. This is

partly because tanks are quickly recharged by

good rains, such as occurred in the monsoon be-

tween the two survey rounds. But it is also because

in the majority of the surveyed villages, the canal

water is fed into the tank and distributed via the

tank’s minor irrigation system. This practice is

very evident in those villages newly connected to

the canal system between the two rounds: the per-

centage of cultivated area receiving

water from a tank jumped from 9 to 53

percent, compared with 0 to 8 percent

receiving water directly from the canal.

South India has a well-developed

T H E P R O J E C T S

1 5

The rehabilitated worksfor SRSP successfullyincreased the irrigatedarea.

The area covered may stillbe below what wasexpected because offlawed estimates.

Figure 2.1: Irrigated Area in SRSP (000s of ha)

0

50

100

150

200

250

300

1995–96 1996–97 1997–98

Are

a

1998–99 1999–2000

KharifRabi

Source: SRSP baseline.

Notes: No water was released in rabi 1997–98. ha = hectares; SRSP = Sriramasager Project.

system of tank irrigation, with some tanks dating

back centuries; existing tanks are sometimes in-

corporated into the canal system. However, as dis-

cussed below, the storage of canal water in tanks

has implications for water allocation and thus can

be contentious.

SRBCIn SRBC, although the main canals and distributaries

were completed, the minor works were only about

20 percent complete as of June 2005, one year

after the project closed. The state government is

completing the works with its own funds, esti-

mating that the works will be finished by 2007. In

2005, some test water had been released, and farm-

ers had breached the main canal to ir-

rigate their fields. Officially, only 7,880

ha had been irrigated, compared with

the planned amount of 65,000 ha.

As noted in chapter 1, the IEG study originally in-

tended to undertake a survey in the SRBC com-

mand based on information from I&CADD (also

given in the Bank’s Implementation Completion

Report), that 44,000 ha had already been irri-

gated. This study found that the 44,000 ha refers

to irrigation from all sources. Counting all irriga-

tion in the command area, whether sourced by the

canal or not, is apparently common practice

(Jairath 2000) and contributes to the overesti-

mation of irrigated area and consequent overes-

timation of benefits and underestimation of the

true per hectare costs of irrigation through major

irrigation schemes.9

These delays seriously affect the project’s viabil-

ity, even if it eventually reaches its full potential.

The Bank’s Staff Appraisal Report (SAR) estimated

the economic rate of return (ERR) for the SRBC

subcomponent to be 13.7 percent. Assuming the

benefits are as stated in the SAR but subject to

the delays that have in fact taken place, this return

drops to 4.1 percent and becomes negative (–1.0

percent) once the original investment costs under

AP II are taken into account.10

SummaryThe AP II project was to newly irrigate 65,000 ha

under SRBC and 163,000 ha under SRSP, as well as

to modernize the system serving a further 165,000

ha under SRSP. However, problems in implemen-

tation meant that no works were finished, so no

farmers received irrigation as a result of the proj-

ect. With positive costs and zero benefits, the rate

of return to the project is negative infinity.

AP III was to complete the works started by AP

II. At the time of this study, the bulk of the con-

1 6


Delays in the completionof SRBC have undermined

project viability.

Figure 2.2: Irrigation Sources, 2005 and 2006

0

10

20

30

40

50

60

Canal 2005NoneTubewellBoreholeTankCanal

2005 2006

Canal 2006 Tank 2005 Tank 2006 Tube 2005 Tube 2006

Perc

ent o

f cul

tivat

ed a

rea

Source Source

0

10

20

30

40

50

60

Perc

ent o

f cul

tivat

ed a

rea

None New Existing

Source: Appendix table D.2.

nections in the SRBC command had yet to be

made. They are currently expected to be com-

pleted by June 2007, nearly 20 years after the

work started.11 Such serious delays, which per-

vade irrigation investments across India and else-

where,12 seriously undermine the viability of the

investment in the SRBC command, which will

be negative even if all intended benefits are even-

tually realized.

Progress has been made in AP III, where the in-

creased carrying capacity of the main canal has

made possible a further extension of the system.

However, the areas irrigated have not reached

those planned at appraisal, and, even immedi-

ately after rehabilitation, actual water flow in the

main canal was 90 percent of capacity (and less

in the distributaries) as a result of siltation, land

slippages, and damaged linings.

T H E P R O J E C T S

1 7

Chapter 3Evaluation Highlights• PIM is classified as pro-poor be-

cause it involves the poor in irrigationmanagement and raises agriculturalincomes and employment.

• Both membership and leadership ofWUAs are biased toward those whoare better off.

• The design of the irrigation systemand management by the WUAs havein some cases adversely affectedwater distribution.

• Only a small percentage of non-WUAmembers believe water is fairlydistributed.

• The reforms have not ended the cycleof poor irrigation services and un-willingness to pay for O&M.

Building a canal for irrigation in India. Photo by Ray Witlin, courtesy of World Bank Photo Library.

2 1

Water User Associations

Origins of PIM in Andhra PradeshA first attempt to change this in India came in 1973

with the initiation of CAD, which relied on farm-

ers’ organizations for water management and

O&M below the distributary level. During the

1970s and 1980s, the Indian government pro-

moted CAD across India, with Bank support,

through the National Water Management Pro-

gram. As today, Andhra Pradesh was a leading

state in those reforms, with the former commis-

sioner for SRSP successfully lobbying for the

creation of a separate CAD Department and be-

coming its secretary for its first seven years (Siva-

mohan and Scott 2005).

These early experiences offer two lessons. First,

in Andhra Pradesh, more than 3,000 farmers’ or-

ganizations were formed by 1981; they soon dis-

banded, because of insufficient funding and an

inability to solve water distribution issues. The

same problems affect WUAs today.

Second, most reforms are based on earlier policy

changes, which often lay the foundation for later,

more far-reaching, changes. Indeed, despite the

failure of the farmer group aspect of CAD, lobbying

for the approach continued, leading to the pro-

motion of WUAs since 1997.

The Current Wave of PIMThe World Bank supported the promotion of

WUAs through AP III and then across the state

through the Andhra Pradesh Economic Restruc-

turing Project (APERP), which included training

in PIM for WUA and government officials. More

than 10,000 WUAs were created following the

passage of the APFMIS Act, which devolved pow-

ers for system management from I&CADD to

farmers.

Specifically, farmers were to assist in collecting

water fees and would receive 90 percent of the

funds collected to undertake O&M.2 WUAs were

also to determine local water distribution through

the warabandi statement (a schedule of water re-

leases). In principle, farmers were to control water

releases throughout the system through the three-

part committee structure (WUA, distributary com-

mittee, and project committee), but this has not

happened.

The reforms were not without oppo-

nents. Although there was strong sup-

port from the top of both the state

government and I&CADD, lower levels

of the bureaucracy were opposed—one reason

project committees have not yet been created.

Under the changes, the lower-level officials known

as lascars (“gate keepers”), each of whom is re-

sponsible for canal operation and water man-

agement across an area of approximately 800 ha,

were to become accountable to the WUAs. This

proposal was challenged in the courts by 3,500 las-

cars across the state (Raju 2001). Although efforts

were made to build cross-party support, the op-

position accused the government of deferring to

User management of irrigation is not a new idea. Indeed, users tradi-

tionally have managed their own water resources through indige-

nous institutions.1 However, the growth of large-scale schemes financed

by the government shifted management responsibilities to the irrigation

bureaucracy.

A 1997 act devolvedirrigation systemmanagement to farmers.

donor agencies by accepting the conditions of the

World Bank, which included the promotion of

PIM (Sivamohan and Scott 2005).3

This move was expected to increase system effi-

ciency. Many supporters of PIM also claim that it

makes irrigation more equitable, ensuring a

greater spread of benefits to the poor—though

that was not an explicit objective in ei-

ther of the two projects under review.4

PIM is claimed to be pro-poor for two

reasons. First, the involvement of the

poor in irrigation management through

WUAs ensures that irrigation benefits are not cap-

tured solely by those who are better off. Second,

the improved functioning of the irrigation system

under PIM results in higher agricultural income

and employment. This chapter reviews the state

of WUAs nearly a decade after their creation by the

APFMIS Act.

WUA MembershipAccording to the Act, all water users are members

of the WUA; “water users” are defined as those

using water from a government or Andhra Pradesh

Water Corporation source.5 The Act states that cul-

tivating households using such water for any pur-

pose, not just agriculture, are to be members.

However, when asked in June 2005 if they or any-

one in the household was a WUA member, just 10

percent of respondents in cultivating households

replied positively.

This low rate is explained by two factors. First, al-

though virtually all plots are irrigated in one way

or another,6 a minority use canals and tanks for ir-

rigation. Membership in WUAs is notably higher

among the latter: in 2006, half of such households

were in WUAs, compared with less than 10 percent

of households using other forms of irrigation (fig-

ure 3.1). Second, many WUAs had ceased to func-

tion because political problems had delayed new

elections7 and because the lack of water meant that

whether the WUA functioned may be thought ir-

relevant.8 But new elections were held in some vil-

lages between the survey rounds, and water was

released along the length of the SRSP command

in July–August 2005. By March 2006 the propor-

tion of cultivating households claiming WUA mem-

bership had risen to 19 percent.

The second factor behind low participation rates

is that, although all households using the water

resources are meant to be members, households

using canal or tank irrigation are most likely to join

a WUA. The bulk of the increase in membership

has come from tank WUAs, reflecting the fact that

tank irrigation has been revitalized as heavy rain

and canal water have recharged tanks in the sur-

vey area.

The fact that not all farm households participate

in WUAs raises questions about the pattern of

participation. Claims that WUAs ensure the par-

ticipation of the poor in irrigation management

must first carry the caveat that the landless, who

make up the bulk of the poorest, are excluded.

This practice contrasts with the approach adopted

in the Andhra Pradesh Rural Livelihoods Project,

supported by the U.K. Department for Interna-

tional Development. In the Rural Livelihoods Proj-

ect, receipt of funds for watershed development

activities is conditional on first implementing

schemes to benefit the landless and marginal

farmers. Moreover, in principle, these latter groups

2 2


Only 10 percent ofcultivating households

were members of WUAs.

Figure 3.1: Households Accessing Canal and TankIrrigation Systems Most Likely to Be in a WUA

0

10

20

30

40

50

60

70

Canal Tank Other irrigation

Prop

ortio

n of

hou

seho

lds

in W

UA

Source of irrigation

2005

2006

Source: IEG survey.

are meant to lead the watershed planning process,

though this rarely happens in practice.

Alternatively, as has happened in other countries,

irrigation projects may include subprojects, such

as fish ponds, that allow the landless to indirectly

benefit from irrigation. Finally, tradable water

rights can be allocated to all households in the

catchment area—strategies to develop this ap-

proach are part of the Bank’s most recent guide-

lines for lending to the irrigation sector in India.

But the bias against the poorest is only partly a re-

sult of the exclusion of the landless. Figure 3.2

shows the proportion of each wealth quartile that

holds membership in a WUA.9 Less than 1 percent

of households in the poorest quartile hold WUA

membership, compared with 28 percent of the top

quartile. Restricting the sample to households in-

volved in cultivation reduces the bias, but not by

much: only 1 in 50 poor households engaged in

cultivation is in a WUA, compared with more than

1 in 3 of the better-off households.

Despite the apparent neutrality of the Act with

respect to caste, landholding, and so forth, there

are also biases in the pattern with respect to

other classifications (appendix table D.4). These

biases are more marked if all households are

considered rather than just farm households,

but they are still present among the latter. Those

households with no education are least likely to

be in a WUA. Large landowners are most likely to

be members, though marginal farmers are more

likely to join than small farmers (figure 3.3).

Households from the top castes are most likely

to be in WUAs and those from scheduled tribes

least likely.10 Using multivariate analysis, the most

robust determinants of WUA membership are

wealth and education (appendix table B.1).

Two factors explain the biases noted. Ownership

of land with access to canal and tank irrigation

is more heavily concentrated among the better off

than is overall land ownership (figure 3.4).11 In

2006, the Gini coefficient for overall land owner-

ship was 0.34, but it was 0.45 for ownership of land

with access to canal or tank irrigation (appendix

tables D.4 and D.5). The second reason behind

the bias is that WUAs operate within the

existing social context, influencing par-

ticipation in ways that often benefit

those who are better off. Such find-

ings are hardly surprising because they

are a well-documented aspect of rural develop-

ment programs in India. The surprise is that more

WAT E R U S E R A S S O C I AT I O N S

2 3

WUAs exclude the poorestfrom decision makingbecause most of them arelandless.

Figure 3.3: Large Landowners Most Likely to BeWUA Members; Marginal Farmers also Active

0Landless Marginal

Amount of land owned

Prop

ortio

n of

hou

seho

lds

with

WU

A m

embe

r

Small Large

5

10

15

20

25

30

Source: IEG survey.

Figure 3.2: Likelihood of WUA MembershipIncreases Sharply with Wealth

01 2 3 4

5

10

15

20

25

30

35

40

All households

Prop

ortio

n in

WU

A

Wealth quartile

Cultivating households only

Source: IEG survey.

was not done to address these known

biases. A better understanding of the

social context in which WUAs operate

would help inform policy so that the

objectives of reform are achieved.

What Do WUAs Do and Who Does It?The APFMIS Act gives WUAs wide-ranging powers,

as they are intended to manage the irrigation sys-

tems.12 However, large parts of these powers have

not been realized, with the Irrigation Department

retaining functions intended to be devolved to the

WUAs. For example, the WUAs are not involved

in revenue collection. Furthermore, funds allo-

cated to WUAs, financed by APERP, were used on

renovation works, with little or none being used

on group development activities.

The Irrigation Department retains control of sanc-

tioning works supposedly under the control of

WUAs; the availability of rents in the contracting

process (Wade 1988), as well as bureaucratic in-

ertia, limit chances that the Irrigation Depart-

ment will give up its own power unless there is a

major political push from the top—an opportu-

nity that was lost once the momentum behind the

reforms faltered. The failure to establish the proj-

ect committees reflects this unwillingness to re-

linquish control.

Water allocationWUAs do have a role in water allocation at the local

level, a process that is frequently marked by rivalry,

both within and between communities. In a ma-

jority of the communities covered by the IEG sur-

vey, most of the irrigation water from the canal is

being channeled into tanks. Tanks have been im-

portant in irrigation in South Asia for centuries,

and incorporating them into the canal system is

a sensible use of existing water distribution chan-

nels. Tanks are also included in the system as bal-

ancing reservoirs (that is, to store water until it is

time for release).

However, evidence from IEG’s field work shows

that storing water in tanks can adversely affect

water distribution. The practice excludes farmers

outside the tank irrigation scheme and some-

times prevents water from reaching those farther

down the canal (see box 3.1): 3 of the 16 vil-

lages surveyed are in the tail reaches of their re-

spective distributaries—none of these villages is

receiving canal water. Another village is not re-

ceiving canal water because the Irrigation De-

partment is opposed to farmers’ requests to direct

the water to the village tank.13 Those farmers not

participating in WUAs—those who are not among

the better off—are most likely those excluded

from benefits, including the half of canal and tank

users who do not participate in WUA activities.

Participation and managementVirtually all WUA members participated in WUA

elections, although this is a minority of those el-

igible to vote, because many eligible households

do not participate in the WUA (table 3.1). One-

third of members regularly take part in meet-

ings;14 somewhat more have participated in O&M.

A smaller number, approximately 17 percent, say

they take part in WUA decision making or be-

lieve they influence water management. This

translates into just 3 percent of cultivators feeling

that they have a say.

Participation in these various activities is not uni-

form across members, but varies according to

the same biases that affect membership. With re-

spect to attending meetings, either at all or reg-

ularly, the multivariate analysis found that the

2 4


Figure 3.4: Lorenz Curve for Land Ownership byIrrigation Source

00 10 20 30 40 50 60 70 80 90 100

10

20

30

40

50

60

70

80

90

100

Cum

ulat

ive

shar

e of

land

ow

ners

hip

Cumulative share of households

Canal and tank

Tubewell and borehole

All land

45°

Source: IEG survey.

Note: Households are ranked by a wealth index.

It is not surprising that thebetter off are more involved,

but it is surprising that theprojects did so little to

address the issue.

uneducated are least likely to participate (ap-

pendix table B.2). When it comes to taking part

in decisions or influencing water management, the

wealth variables are significant; indeed, no one in

the bottom quartile replied positively with re-

spect to either of these questions (appendix table

D.7). For voting, as this is near universal, there are

no significant biases among members; again,

members are the minority.

Differential influence is reflected in the pat-

tern of WUA leadership, which is dominated by the

better off. In a survey of 222 WUAs in 22 districts,

88 percent of leaders of canal system WUAs were

from higher castes and 11 percent from backward

castes; just 1 percent were from scheduled tribes

or scheduled castes. Large landowners accounted

for 88 percent of canal WUA presidents.

Elite domination is less marked for tank systems

but still present: 86 percent of presidents are

from higher castes or backward castes and 56

percent are large landowners (Reddy and Reddy

2005). Membership in the WUA committee is

more representative (for example, ORG 2005), but

in many cases the president runs the WUA with

little consultation: in a survey of 214 WUAs, the

majority was not meeting regularly (ORG 2005).15

Irrigation Department officials say that canal design

is determined technically and so cannot be subject

to political interference. Design of the micro-

network ideally is done in consultation

with local communities, but this is not

what happens in practice. Indeed, in one

IEG study village, politically well-connected

families are not among those who will

benefit and are mobilizing people against

the completion of the canal and obstructing

land sales. Nonetheless, there is substan-

tial scope for extracting more water than

envisaged in the system’s design.

The impact of WUA management on water

allocation was captured in a question that

asked if the respondent believed water

to be fairly distributed within the com-

munity. Only 15 percent of non-WUA mem-

bers believe that water is fairly distributed,

compared with 55 percent of mem-

bers, suggesting that WUA members

are able to enjoy better access than

nonmembers. In a multivariate analy-

sis of the response, only one socio-

economic characteristic is significant:

the larger a person’s landholding is,

the more likely he or she is to believe that water

is fairly distributed (figure 3.5; regression results

in appendix table B.3).16

The finding that WUA members are significantly

more likely to say that water is fairly distributed

than nonmembers is robust. In contrast, the pres-

ence of an active WUA in the village (measured by

the village-level WUA membership rate) has at

best no impact on perceived fairness. In some

model specifications it increases the likelihood that


2 5

Box 3.1: Diversion of Water to TanksLimits the Number of Beneficiaries

In one village in Warangal district, the canal water isdiverted to a tank, depriving two other villages fartherdown the canal of water. This has caused clashes be-tween the villages.

In another village, the canal water flows into fivetanks, each of which has a functioning WUA. There isalso a canal WUA, but it is not operational.

Sources: IEG field visits.

Because much of theirrigation water is storedin tanks managed byWUAs, water distributionhas been adverselyaffected.

Table 3.1: Participation in WUA Activities, 2006 (percent)

All Households with WUAcultivating access to canal membershouseholds or tank irrigation only

Attend meeting 10.8 30.2 58.3

Attend at least six meetingsa year 5.8 15.1 31.0

Participate in O&M 7.3 20.5 39.0

Take part in decision making 3.3 9.1 16.8

Influence water management 3.1 9.4 17.6

Voted for WUA president 17.9 49.0 96.1

Voted for committee chair 17.1 46.5 91.7

Number 557 150 99Source: IEG survey.

people will say that water is not fairly

distributed. The explanation for this

finding is clear. Being in a WUA pro-

vides access to water, which is shared

among WUA members. This process

excludes nonmembers, who thus feel

that water is not being fairly shared.

Hence, the more active the WUA, the

more likely nonmembers are to feel

that water is not shared fairly.

Although it is expected to ensure eq-

uitable water distribution, the WUA

lacks the powers of enforcement needed to do

this.17 Field work found that canal breaches are

common practice, meaning that excess water is

taken off in higher reaches, leaving tail-enders

without enough water. Such matters should be

broached in the WUA and distributary committee

and sometimes are. But the farmers’ groups have

no mechanism for resolving the dispute. Those

breaching the canal may be connected to powerful

families, making it difficult to challenge them in

an open forum. In one surveyed village, the Irri-

gation Department is taking the farmers to court

for breaches.

Higher reaches also use more water than in-

tended as a result of the preponderance of paddy

cultivation (see next chapter). They can also cap-

ture a larger amount of water by diverting the

canal into their village tanks. Of the 18 villages in

the IEG survey, one was not receiving water at all

because of diversion upstream, and two others

were diverting water and were therefore in con-

flict with downstream villages that were not re-

ceiving water.

The distributary committee should resolve these

disputes but cannot.18 Such problems are a man-

ifestation of the fact that decentralization is of-

ten best accompanied by an element of central

control.19

O&MWUAs can claim more success with O&M, al-

though questions might be raised regarding its

sustainability. In SRSP, WUAs were involved in

identifying repair works, though Irrigation De-

partment staff are responsible for repair design.

This approach was adopted statewide in the

late 1990s, partly using World Bank financing.

Most studies are positive about the works carried

out at that time (Jairath 2000; Reddy and Reddy

2005).

It is argued that the involvement of the WUAs in

identifying the works ensured that priority activ-

ities were undertaken and forced a change in

I&CADD’s preferred way of carrying out mainte-

nance on minor canals (Oblitas and others 1999).

One consequence of this was that minor canals

would be in a very poor state before they would

be repaired. Moreover, it is also claimed that cost

savings of about 20 percent were realized as a re-

sult of avoiding rent-seeking by public officials at

various levels (Raju 2001).20

A further cost saving is that the assumption of

these responsibilities by WUAs means that

I&CADD can operate with lower staffing levels.

However, the decisions were usually made by the

WUA president, with little or no consultation with

other committee members, let alone ordinary

WUA members.21 Moreover, WUA presidents are

2 6


Only 15 percent of non-WUA members

believe water is fairlydistributed, compared

with 55 percent ofmembers.

WUAs lack the powerthey need to enforce

equitable waterdistribution.

Figure 3.5: Larger Landholders More Likely to ThinkWater Is Fairly Distributed, as Are WUA Members

Like

lihoo

d of

say

ing

wat

er d

istr

ibut

ion

is fa

ir

Land owned (ha)

0.00 5 10 15 20

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

WUA member

Non-WUA member

WUA effect

Source:

themselves frequently contractors: in one study

of eight WUAs, at least six presidents—and pos-

sibly all eight—were contractors who undertook

canal rehabilitation.22

Although it is too early to make a fully informed

judgment, it seems doubtful that WUAs will be ef-

fective in carrying out O&M because the financ-

ing system is unsustainable. Following the reforms,

the government allocated Rs 200 per acre to major

schemes. This figure balances with the water fee

of Rs 200.

But the apparent balance gives a false picture

of financial sustainability for several reasons.

First, 10 percent of the fee revenue goes to local

government rather than any of the water com-

mittees. Second, in major schemes, 20 percent is

allocated to project committees, but there are

no such committees and these funds are not

being paid out.23 Most important, collection re-

mains poor. The amount collected has risen some-

what as a result of tariff increases, but this probably

disguises an actual reduction in collection rates.

The reforms have failed to break the vicious cycle

of poor irrigation services, making farmers unwill-

ing to pay for O&M. Specifically, tail-enders continue

to receive little or no water and are thus unwilling

to pay the water fee. But even farmers in head

reaches frequently do not pay. Collection rates

have been estimated at less than 10 percent of the

amount due (Jairath 2000); data from the SRSP

command show an average of 12 percent for the

period 1998–99 to 2001–02 (I&CADD 2004). Hence,

over the period 1997–2002 only one-third of the

money allocated to canal WUAs and distributary

committees was financed from water charges.24

Moreover, less than half of WUA members believe

that repair works are a WUA responsibility (Reddy

and Reddy 2005). For that reason, prospects are

not good for WUAs sustaining O&M so that the

canal network does not fall victim to the same

problems of poor maintenance. The positive coun-

terpoint is that (as is shown in the next chapter)

in good years, farm income is easily sufficient to

cover even higher water charges.

SummaryThe APFMIS Act was intended to pro-

vide the basis for transferring manage-

ment of the irrigation system from

public servants to farmers. The gov-

ernment at the time was heavily com-

mitted to the reform and provided

support to WUA formation. But reforms

lost momentum with the change of

government, and the transfer has not

taken place. The Irrigation Department

has retained overall control of the command areas

because project committees have not been formed.

WUAs are not involved in revenue collection, and

most farmers see O&M as a government respon-

sibility rather than one for themselves or the

WUAs.

Supporters of PIM claim that it is a channel for rep-

resenting the voice of the poor. This has not been

the case for several reasons. First, the poorest are

excluded because they do not themselves engage

in cultivation. But even among farm households,

a tiny minority of poor households participate in

WUAs, compared with a sizable proportion of the

better off.

Among WUA members there are also biases in the

pattern of participation. The uneducated are less

likely to participate in meetings, and the better off

are most likely to feel they have a role in decision

making and influencing water management. Lead-

ership of WUAs is heavily concentrated among

higher castes. For these reasons, WUAs can be-

come a channel for elite capture rather than for

empowerment of the poor. Evidence from the

field supports the argument that WUAs can be-

come hostage to local politics, which may mean

elite capture but can also mean capture of proj-

ect benefits by one group or locale at the ex-

pense of others.

Water is not being fairly distributed either within

or between communities, and WUAs have only

limited ability to do anything about this. Many of

these findings reflect similar findings for other

community-based initiatives supported by the

Bank.


2 7

WUA involvement in O&Mhas shifted to addresspriority problems.

The financial sustain-ability of O&M is doubtfuland may undermine WUAsuccess in that area.

In summary, the Bank has been overly optimistic

about what may be achieved through WUAs ( this

supports the conclusion of IEG’s sector study,

IEG 2002, pp. 43–45), underestimating the need

for continued government involvement. This ap-

plies to technical issues, such as the design of

rehabilitation works, as well as to the ability to

intervene in water disputes. For the time being at

least, it also applies to continued subsidization of

system operation. At the same time, the reasons

for decentralizing WUAs need to be remembered,

in particular, the weaknesses of government irri-

gation departments in efficient management of the

irrigation system. Successful reform needs to ad-

dress these weaknesses with a realistic assess-

ment of what WUAs can and cannot do.

2 8


Chapter 4Evaluation Highlights• The impacts of irrigation on crop

yields were strong, but not as signif-icant as appraisal estimates putthem.

• The impact of irrigation on croppingdiversity was minimal, overridden bya strong preference for cultivatingrice.

• Cropped area and cropping intensityincreased, though not as much asappraisal estimates predicted.

• Irrigation increased farm incomesand smoothed income from year toyear.

• Employment of nonhousehold laborincreased with irrigation, benefitingwomen the most.

• The ERR to investments in these proj-ects was negative.

• Risk calculations in Bank appraisalsare too simplistic.

Two men working in an irrigated field in India. Photo by Ray Witlin, courtesy of World Bank Photo Library.

3 1

Economic Benefits

Direct Production ImpactsThese production impacts come from three

sources: higher yields, diversification into higher-

value crops, and increased cropping intensity.2

Higher yieldsAt appraisal, the main income benefits of irrigation

were expected to come from higher yields, which

would result in higher net farm income. For ex-

ample, in the SRSP command below Lower Man-

air Dam (which includes the areas covered by the

survey), the SAR predicted that paddy yields would

rise from 3.2 metric tons per ha without the proj-

ect to 5.4 metric tons with it.3 That is, the SAR as-

sumed a 69 percent increase in paddy yields. This

percentage increase is toward the upper end of im-

pact estimates, with the likely value somewhat

less. But it is also important to note that the “with-

out project” yield is high given actual yields.4 This

matters because it is the absolute increase in yields

that determines the absolute increase in net farm

income, which is the main project benefit.

This study used several sources to estimate the

impact on yields. The IEG survey data were used

to calculate double-difference estimates (see ap-

pendix C).5 These data came only from the reach

below Lower Manair Dam, so additional data from

the Implementation Completion Report (ICR),

SRSP baseline data, and data collected by IEG

from various official sources in Andhra Pradesh

were used to make impact estimates for the higher

reaches,6 though these estimates are single dif-

ferences. These two approaches give consistent

results of an increase in paddy yields of 1–1.5

tons per ha, that is, around 20 percent–50 per-

cent.7 The yield increases for turmeric and chilies

are less in absolute terms but in a similar range

in percentage terms. Cotton has a larger per-

centage increase of around 80 percent and is the

only crop for which the impact on yields appears

to come close to that assumed at appraisal.

An alternative approach is to simply

compare crop yields for the same year

on plots without irrigation with those

with access to canal or tank water. This

analysis was done for paddy in 2006.8

The results show that the impact of ir-

rigation on yield was relatively small in

the 2005–06 kharif season (which is to be ex-

pected, because the monsoon was good) but

larger in the rabi season. Averaging over these

results gives a similar result to that found in the

double-difference analysis.9 Finally, the “natural

experiment” of comparing the year in which

no water was released in SRSP with earlier years

gives an impact on rice yields of just 14 percent

in the kharif season but 66 percent in the rabi

season.

The analysis offers two clear findings. First, access

to irrigation has a significant impact on yields

(figure 4.1). But, second, these increases are less

than those assumed at appraisal, ranging from

one-third for the high-value chilies and turmeric

(34 and 37 percent, respectively) to just under

three-quarters (72 percent) for cotton. For paddy,

Direct production benefits accrue to farm households accessing irri-

gation. Once the SRBC project is completed, the investments un-

dertaken by AP II and AP III will provide these direct benefits to

approximately 212,000 households.1

Estimates of impact onyield found a consistent20 percent–50 percentincrease—substantial, butwell below appraisalestimates.

the most important crop, the assumed

impact at appraisal was almost twice

what has actually been realized.10 Com-

pared with appraisal expectations, these

shortfalls in yields are the norm rather

than the exception in irrigation proj-

ects.11 Because no data are available for the SRBC

command (few farmers have been connected to

the canal system), similar reductions in expected

yield are applied to that case (see appendix C).

Assuming everything else remained as assumed

in the appraisal report, then halving yields com-

pared with their expected values reduces the ERR

in the SRSP (below Lower Manair Dam) area from

the 21 percent calculated at appraisal to 12 per-

cent.12 Applying the same 50 percent reduction

in yields to the whole project reduces the over-

all ERR from 24 percent to 8 percent.13

DiversificationIrrigation is intended to promote agricultural di-

versification, which helps reduce poverty by in-

troducing higher-value crops. Also, a diversified

production portfolio is more resilient to nega-

tive external shocks. The economic analysis in

the SARs for the two projects assumes a diversi-

fied production structure on farms receiving canal

irrigation, notably the production of a broader

range of high-value cash crops, although it is

stated that agricultural support services are nec-

essary to attain diversification.

However, the survey data show that households

with irrigated plots have less diversified cropping

patterns, and farms with access to canal and tank

irrigation are less diversified than those depend-

ing on boreholes and tubewells. The lower level

of diversification reflects the greater reliance on

rice by those with access to irrigation schemes.

These findings can be shown in several ways.

Figure 4.2 shows the share of cropped area, clas-

sifying households by the type of irrigation to

which they have access. The figure also shows the

crop shares that the Bank assumed would be

adopted in project areas, with clearly consider-

ably more diversification being assumed than

has been realized. Those with tank and canal ir-

rigation devote the highest share (55 percent of

gross cropped area) to paddy, 6 percent more

than do farmers without irrigation. The share of

rice in total crop income in 2006 was about 15 per-

cent greater for farmers with canal or tank irri-

gation and more than 20 percent higher than

the year before (appendix table D.9).14 Single-

difference estimates using the IEG data put the

share of paddy in areas with canal irrigation at

up to 30 percent more than in other areas, and

double-difference estimates suggest this increase

is 10 percent to 20 percent. Data from the ICR

show the share of paddy in gross cropped area

up to 39 percent higher in irrigated areas than in

areas without irrigation.

These figures partly reflect differences in the

share of households that depend exclusively on

rice. Compared with other households, twice as

many households with access to canal and tank

irrigation cultivate paddy exclusively (47 percent

in 2005 and 37 percent in 2006). Other than rice,

canal- and tank-irrigated farmers mostly grow cot-

ton; they do not diversify into groundnut, maize,

or high-value crops.

The lack of diversification is not an especially sur-

prising outcome because Andhra Pradesh farm-

3 2


Figure 4.1: Irrigation Increases Yields . . . But by Lessthan Assumed at Appraisal (change in yield, tons/ha)

0.0

0.5

1.0

1.5

2.0

2.5

Paddy Cotton Chilies Turmeric

Chan

ge in

yie

ld (t

onne

s/ha

)

Appraisal estimate

Actual impact

Source: IEG survey.

Households with irrigatedplots were found to haveless diversified cropping

patterns.

ers are known to have a strong preference for cul-

tivating rice. Explanations for this preference

range from its reputation as an “easy crop” to

food security and the fact that it enjoys a gov-

ernment-supported price, whereas other crops

may experience much more price volatility.15

Moreover, if a farmer chooses a wet crop, espe-

cially paddy, then farmers in neighboring fields

have little choice but to follow suit, as water leak-

age means their own soil will be unsuitable for

other crops. What is surprising is not the lack of

diversification, but the extent to which it was be-

lieved at appraisal that farmers would diversify de-

spite strong evidence to the contrary. That

evidence has included repeated failure of the

government to enforce production of less water-

intensive crops in the upper reaches of irrigation

systems. The failure to diversify is unique nei-

ther to this project nor to Andhra Pradesh (see,

for example, Plusquellec 2002).

Diversification was expected to result from ex-

tension services, but the data show that few farm-

ers have received such services (ap-

pendix table D.10). The main area of

advice has concerned seeds and use of

fertilizer, but this has still reached only

about 7 percent of households and has

come mostly from community workers (most

likely through the Restoration of Livelihoods Pro-

gram or the Indira Kranti Patham Program) rather

than government extension officers. Construc-

tion delays meant that the support the projects

gave to extension services took place long be-

fore farmers received water, so that support would

have been of limited relevance. But even then such

services were limited: the intended small budget

share for agricultural support services at just 2 per-

cent was squeezed out, falling to just 0.6 percent.

The claims that effective extension ser-

vices help support higher yields and di-

versification may be true, but this

cannot be tested in the virtual absence

of such services in the areas surveyed.

The lack of diversification may not matter much

to those receiving a direct income benefit from

E C O N O M I C B E N E F I T S

3 3

Figure 4.2: Cropping Pattern by Irrigation Source, 2006 (share of gross cropped area)

0

20

40

60

80

100

Canal and tank

Borehole Tubewell Other SAR "with project”

Irrigation Source

New canal New tank

Shar

e of

gro

ss c

ropp

ed a

rea

(%)

Paddy Maize Groundnut Chilies and turmeric Other cropsCotton

Source: IEG survey.

The data confirm a knownstrong preference amongAndhra Pradesh farmersfor cultivating rice.

Although extensionservices were intended tohelp, few farmers receivedsuch services.

irrigation, because rice generates a reasonable

and safe net income and provides food security.

But the additional water requirements of paddy

production have adverse implications for two

other groups of cultivators, most likely to be the

poorer cultivators.

First, the system is designed to provide adequate

water for the cultivation of irrigated dry crops,

such as groundnuts, cotton, and maize. Areas

can be declared as reserved for irrigated dry

crops. However, as the results show, the policy

is not enforced. Canal water charges are on an

area basis rather than volumetric basis, and col-

lection of even those charges is weak, so farm-

ers have no incentive to change cropping

patterns. Indeed, rice is a high-margin crop, pro-

viding a more stable income because the gov-

ernment guarantees the price. Hence, the lower

reaches of the command will only re-

ceive water in very good years, result-

ing in the shortages and disputes

described in the previous chapter and

reducing project benefits.

The second group of losers comprises house-

holds dependent on rain-fed agriculture. The fail-

ure of irrigated farmers to switch to high-value

cash crops means they remain as competitors to

rain-fed farmers, who suffer from lower prices as

a result. But landless households and marginal

farmers who are net food buyers benefit from

these lower food prices. Within the survey area,

just over 40 percent of households (including

noncultivators) do not grow rice at all; another 10

percent grow insufficient amounts to meet their

own consumption needs and so are net buyers.

Thus, half the households will gain from the

irrigation-induced price reduction, although the

result might instead be fiscal savings.16

Cropped area and cropping intensityIrrigation increases the area planted

with crops both by allowing cultiva-

tion of previously uncultivated areas

and by increasing cropping intensity.

Cropping intensity can be increased

by adding growing seasons or by multicropping

(planting more than one crop on the same plot

at the same time, although this is not common on

irrigated land). The area of actual land cultivated

is the net cultivated area; gross cultivated area

refers to the total with double counting (or more)

of land that is multicropped or cultivated in more

than one season.

The net area cropped by the households sur-

veyed increased by nearly 10 percent between

2005 and 2006; cropping intensity increased by

nearly 4 percent, adding up to a 13 percent in-

crease in gross cropped area (appendix table

D.11). Different approaches to assessing project

impact in the SRSP area give estimates of an in-

crease in cropping intensity from 10 percent to

23 percent (appendix table C.18).17

The increase in cropping intensity comes from

both greater intensity of individual irrigation

sources and the adoption of new irrigation sources

that have higher cropping intensity. The main

shift is away from unirrigated plots with low crop-

ping intensity toward higher-intensity tank irri-

gation (figure 4.3).

Canal and tank irrigation are associated with crop-

ping intensities of 200 and 150 percent, respec-

tively (appendix table D.12), an increase that

comes from the possibility of a second, or even

third, crop per year. Double-difference estimates

suggest that canal irrigation increases cropping in-

tensity by 40 percent and tank irrigation increases

it by 20 percent. This average increase of around

30 percent is somewhat above the single-

difference estimates from elsewhere in SRSP and

at the upper end from experience elsewhere.

However, it is the number of sources rather than

the specific source that is the primary determinant

of cropping intensity.18

The survey data show that the increase in crop-

ping intensity comes from the increase in the

number of growing seasons rather than multi-

cropping (figure 4.4). The survey found a 17 per-

cent increase in the number of crops being grown

(separate crops reported on the plot-by-plot col-

lection of production data). This increase came en-

tirely from the expansion of production beyond

the kharif season: in 2005, 96 percent of all crops

3 4


The increase in paddycultivation adversely

affects farmers whocultivate other crops.

The increased croppingintensity was from an

increase in the number ofgrowing seasons rather

than from multicropping.

were grown in the kharif season, compared with

71 percent in 2006. There was no increase in mul-

ticropping between the years. Double-difference

estimates suggest that canal irrigation raises the

number of crops grown on a plot by 0.5, and tank

irrigation increases that number by 0.15 (corre-

sponding to increases in cropping intensity of 38

percent and 12 percent, respectively).

Direct Income Effects: Higher Farm IncomeOutput effects are estimated by calculating the

double-difference value of farm output using

2005 prices for both 2005 and 2006 (see appen-

dix C). The results show that access to irrigation

increases the constant price gross value of out-

put, although the estimates span a large range.

Access to canal irrigation is found to increase

gross revenue by 20 percent, and tank irrigation

increases it by 5 percent.

Farm income is affected by the changing level and

composition of output, as well as by changing

input and output prices. In 2005, at the height of

the drought, the rice price was high; it fell be-

tween 2005 and 2006. However, prices of cash

crops rose between survey rounds (see appendix

table D.13). This study does not attempt to attrib-

ute these price changes directly to the project.

In general, however, the increase in paddy pro-

duction, as well as increased labor demand, will

decrease margins, partly offsetting the yield ben-

efits from irrigation. But given the substantial rise

in cash crop prices, notably those of turmeric

and chilies, the double-difference impact of irri-

gation on incomes appears substantial, reaching

75 percent for canal irrigation.

The farm model finds that irrigation increases

net farm income by 110 percent. About half of this

increase comes from increased cropping intensity

and most of the remainder from higher yields, with

only a small part attributable to changes in the

crop mix. However, these increases are smaller

than those anticipated in the SAR, which esti-

mated that income would more than double in

areas where canals were rehabilitated and nearly

triple in newly irrigated areas.

In general, irrigation, especially large-scale

schemes that are not dependent on the local wa-

tershed for water, guarantees water supply and

so smooths annual variations in production. Crop


3 5

Figure 4.3: Cropping Intensity by Number ofIrrigation Sources, 2006

0

40

20

60

80

100

120

140

160

180

200

0 1

Number of irrigation sources

2

Crop

ping

inte

nsity

(%)

Source: IEG survey.

Figure 4.4: Increased Cropping Intensity Comesfrom Additional Growing Seasons

0

100

200

300

400

500

600

700

800

900

Kharif Rabi

2005 2006

Other Kharif Rabi Other

First cropSecond crop

Third crop

Fourth crop

Source: IEG survey.

failures should occur less on irrigated

land than on unirrigated. However, be-

cause of the prolonged drought, in the

2004–05 season no water was released

in the SRSP command for the first time since it

opened. Falling groundwater levels meant that

local irrigation sources were also either dry or that

they provided less water than usual. In 2005, 10

percent of cultivating households got no crop in-

come at all (compared with less than 0.5 percent

in 2006), and the crop failed on 10 percent of all

plots.

The rise of farmer suicides in the state

has been linked to crop failure. In many

instances, farmers could not repay

loans they took to construct a tube-

well, which then ran dry. But unirri-

gated plots were four times more likely to

experience crop failure, except compared with

tank irrigation (table 4.1). In 2006 there were vir-

tually no crop failures on irrigated land. In con-

trast, on unirrigated land, crops failed on 5 percent

of plots. The data do not allow further investiga-

tion of this issue, but evening out crop income is

a substantial benefit from irrigation.

Farm-level returns from irrigationThe returns to the farmer using canal and tank

irrigation are high, as there are substantial bene-

fits and very low costs. The investment costs for

higher yields and cropping intensity are low, as the

farmer is responsible only for the field canals,

which distribute the canal water around the house-

hold’s fields. Moreover, canal water is normally re-

leased into the field by a free intake

(flooding), so most farmers can avoid

pumping costs (equipment and run-

ning costs).19 The recurrent costs as-

sociated with canal and tank irrigation

are the water fees and contribution of

labor to O&M undertaken by the WUA.

Even after the rate increase, the water

fee is only a fraction (less than 3 per-

cent) of the value of the incremental

farm income from gaining access to ir-

rigation. The fee is often not paid, and

few WUAs organize maintenance work.

The farm model (appendix E) shows an average

61 percent increase in net farm income from

canal and tank irrigation.

Tubewell or borewell irrigation also has a high rate

of return, although lower than from canal or tank

irrigation. With this type of irrigation, the farmer

bears the initial investment cost and the recurrent

costs of pump operation.

These costs are lowered by the state govern-

ment’s policy of a zero tariff for agricultural elec-

tricity connections; this is essentially a regressive

subsidy, as it is enjoyed by the better-off farmers.

When it was introduced, the policy could be de-

fended because many farmers were enduring se-

vere economic hardship from years of drought.

But in normal and good years, the benefits from

irrigation easily cover the actual cost of electric-

ity. Unfortunately, the impact of the policy is sim-

ply to encourage overuse of water, to the

detriment of groundwater levels and hence of

other users.

The fact that canal irrigation has sizeable bene-

fits but low costs—and indeed no initial costs to

farmers other than constructing field canals,

which can be done using household labor dur-

ing slack periods—means that there are poten-

tially great benefits for poorer farm households

that cannot access boreholes or tubewells be-

cause of the high initial costs. Those same high

private benefits make the wells attractive to poor

and non-poor alike, so the non-poor will seek

to assert influence to gain access. This occurs in

two ways. First, head-enders use more water than

allowed so that tail-enders get little or no water.

Second, canal water is diverted to tanks so that

only farmers with access to the existing tank sys-

tem get water.

Indirect Income Effects: Employment GenerationThe indirect benefits of irrigation are commonly

referred to as multiplier effects and are often

claimed to be substantial for irrigation.20 These ef-

fects include increased labor supply and demand

for other inputs.

3 6


Irrigation substantiallysmoothed crop income

from year to year.

Well-based irrigation alsohas high returns, though

lower than from canal ortank irrigation.

Table 4.1: Crop Failure by Irrigation Source (% of plots)

Source 2005 2006

Canal 6.1 0.0

Tank 15.6 0.0

Tubewell 5.5 0.0

Borehole 4.9 1.3

None 20.6 5.1Source: IEG survey.

If a macro model, such as a computable general

equilibrium model, is used to calculate these ef-

fects, then all rounds of consumption and pro-

duction multipliers are included. But it is not

clear that these multiplier effects should be in-

cluded in a project-level analysis. Such indirect

benefits exist for all projects, not just irrigation,

and are rarely calculated, certainly not beyond the

second round. Hence, resorting to multiplier ef-

fects to boost the benefit stream will undermine

comparability between irrigation and other in-

vestments. More critically, multiplier calculations

have a risk of double counting. Proper use of

shadow prices is preferred.

Take the example of additional employment. If the

labor demand is fully additional, that is, if the

workers would be otherwise unemployed, then

the appropriate shadow wage rate is zero, and the

wage benefits do get included in the benefit

stream.21 There is a complication only if distrib-

utional weights are being used, as the wage ben-

efit should be weighted by the wage workers’

weight, not that of the farm owner. This report

does not use distributional weights but in this sec-

tion looks at who gets the benefits from irrigation.

The principle spillover benefit of irrigation is ex-

panded employment opportunities for members

of landless and marginal households. Agricultural

employment rose markedly between the two sur-

vey years, from 155 days per household each year

to 223 days (see appendix table D.15). Much of this

increase resulted from the revival of agricultural

fortunes following the 2005 monsoon season.

More detailed analysis is required to determine

how much of the increase can be attributed to

irrigation.

The data show that employment of nonhousehold

labor (from outside the household) increases

with the adoption of irrigation. The double-

difference estimate shows that access to canal ir-

rigation increases employment of nonhousehold

labor by approximately 60–80 days a year. Tank

irrigation appears to increase employment some-

what less, with increases from borehole and tube-

well irrigation falling between the two. It may be

the case that canal irrigation required

additional labor for the construction or

rehabilitation of field canals. Tubewell

and borehole irrigation systems have a

greater number of waterings each sea-

son, which can explain the higher labor demand

with respect to tank irrigation.

It might be expected that the additional labor

demand would be muted by households making

greater use of domestic labor. This is not the

case. The double-difference estimates suggest

that irrigation in fact reduces domestic labor use.

To understand this result, one must understand

the general situation, in which use of domestic

labor fell on average from 336 to 317 days22 be-

tween 2005 and 2006. This decrease may be in part

an income effect, substituting leisure for work as

income rises. The more likely explanation is that

households hire out their labor in response to the

greater labor demand by other households.23

Given the lack of alternative employ-

ment in 2005, domestic labor was ab-

sorbed into the family farm, but not in

an efficient manner: labor productivity

was nearly 20 percent lower in 2005

than in 2006.24 With the good rains improving

agricultural performance in 2006, combined with

new irrigation, there was greater labor demand.

Each household supplied on average 223 days of

agricultural labor in 2006, compared with 155 in

2005. The increase in demand for labor led to an

increase in average wages of 10 percent for women

and 5 percent for men; the greater increase in

women’s wages reflects the disproportionate de-

mand for female labor.25 Women’s wages rose

the most in areas of new irrigation.

Women account for the majority of hired labor

in Indian agriculture, with women’s share typi-

cally ranging between 55 and 66 percent (Upad-

hyay 2004). In the IEG survey, women accounted

for 63 percent of hired agricultural labor (share

of days hired) in 2005 and 64 percent

in 2006 (appendix table D.17). In other

words, close to two-thirds of the em-

ployment benefits accrue to women,


3 7

The lower costs to farmersfor canal irrigation havepotentially great benefitsfor the poor.

The survey found thatemployment of nonhouse-hold labor increased withirrigation.

Women gained most from the increase inemployment.

reflecting reliance on female labor for

key tasks in paddy production, in par-

ticular weeding and harvesting. A con-

siderable body of evidence shows that

women’s income has a larger impact on child wel-

fare (health and education) than does men’s in-

come, even allowing for the fact that engaging

in labor removes the primary caregiver from the

home.

The employment benefits are not just more em-

ployment but its distribution across the year.

There is much migration from areas of rain-fed

agriculture (mainly to urban areas) because of

lack of income opportunities in rural areas. In

2005, because there was little production in the

rabi season, the average employment of out-

side labor during the rabi season was just 2

days. In 2006, this figure rose to 46 days (ap-

pendix table D.16). The impact analysis shows

that the largest part of the employment effect in-

deed comes in the rabi season, thus smoothing

the seasonal fluctuation in employment (figure

4.5). Despite the monsoon in July–August 2005

that brought rain, rabi season agricultural em-

ployment was negligible in villages without irri-

gation in 2006—but significant in villages with

new irrigation. These figures confirm that a large

part of the additional employment is an “irriga-

tion effect,” which reduces the need for sea-

sonal migration.

Who Benefits? Poverty Reduction andDistributional Effects of New IrrigationThe income benefits from irrigation accrue in the

form of higher net income for cultivating house-

holds and wage income for those they employ,

which in turn is made up of higher wages and

increased employment. The farm model (ap-

pendix E) produces figures for both of these

benefits for a typical hectare of irrigated land. To

see how these benefits are distributed, the in-

cremental farm income is allocated among wealth

quartiles according to access to irrigation, the in-

come from incremental employment to share

of the marginal employment, and wage increases

across all employees. The results are shown in

figure 4.6.

The results show that, in absolute terms, the ben-

efits from irrigation are skewed toward those who

are better off. Those in the poorest quartile ben-

3 8


Figure 4.5: Seasonal Employment Effects

0

20

40

60

80

100

120

Canal Kharif2005 2006

In 2005, none of the farms receivedwater, and fluctuation was similar.

In 2006, irrigated farms, which received water, showed less fluctuation.

Rabi Kharif Rabi

Areas withoutirrigation

Areas with new irrigation in 2006

Tank

Add

ition

al e

mpl

oym

ent c

reat

ed (d

ays)

0

5

15

10

30

35

20

25

40

45

Agr

icul

tura

l em

ploy

men

t per

ha

Kharif

Rabi

(a) Largest share of the employment increase from irrigation comes in the rabi season.

(b) Seasonal employment fluctuation is reduced in irrigated areas.

Source: Appendix C.

Notes: (a) is based on household data and (b) on the village-level comparisons.

Irrigation also smoothedseasonal fluctuations in

employment.

efit least from higher net income because few of

them have land to irrigate. Larger farmers are

concentrated in the top quartile, which captures

the bulk of the higher farm income.

This bias is only partially offset by increased em-

ployment opportunities for two reasons. First,

the incremental wages (additional employment

plus higher wages) account for only a minor

share (25 percent) of the total benefits, most of

which are additional employment rather than a

wage increase.26 Second, those in the poorest

quartile benefited little from expanded employ-

ment, which went mainly to households from

the second and third quartiles. The poorest quar-

tile does not benefit from expanded employ-

ment for a number of reasons, including the fact

that it includes elderly and those suffering from

disabilities who are unable to engage in such

work.

Because the benefits to the poorest quartile are

limited, the impact of irrigation on poverty de-

pends a great deal on where the poverty line is

drawn. If it is set at a level such that only those

in the bottom quartile are poor, then the impact

is small (figure 4.7). But with a higher poverty line,

there is a notable impact: bringing irrigation to

a village results in a one-off reduction in poverty

of 10 percentage points, equal to a 25 percent fall.

As discussed below, the dynamic impact on

poverty as a result of reduced vulnerability will

be greater still. There is, however, little impact on

the poverty gap using either poverty line be-

cause of the limited benefits to the poorest.

Although the top quartile benefits most in ab-

solute terms, in relative terms the largest benefit

goes to the second quartile, members of which ex-

perience income growth of 30 percent. Despite

their low benefit, their already low income means

that the poorest quartile also experiences income

growth of 20 percent, compared with 19 percent

growth in the top two quartiles. These effects

mean that irrigation has just a modest positive im-

pact on distribution. However, another major ef-

fect on distribution comes from the dynamic

effects.

Analysis of the Lorenz curves for in-

come between 2005 and 2006 shows a

dramatic reduction in inequality. This

is because many more households had

zero or negative income in 2005 than in 2006

(figure 4.8a). This change was mainly brought

about by the end of the drought. But it was shown

that irrigation reduces the chance of crop fail-

ure, which pushes otherwise well-off families to

the bottom of the income distribution. This fact

is shown in reverse in figure 4.8b, which plots the

Lorenz curve for 2005 and the distribution of in-

come for 2006 with households ranked by 2005

income. The bottom 10 percent of households in

2005 got 10 percent of income in 2006 and 20 per-

cent of farm income (not shown).

So, although comparing “with” versus “without” ir-

rigation for a normal or good year will show a

modest impact on the income distribution, bad

years will be marked by an extreme

worsening in income inequality, as many

farmers experience negative income.

The presence of irrigation can mute this

effect. This reduction in vulnerability


3 9

Figure 4.6: Distribution of Irrigation BenefitsSkewed toward the Better Off

01 2 3 4

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

Wealth quintile

Inco

me

gain

per

hou

seho

ld (R

upee

s)

Incremental wages

Incremental employment

Incremental farm income

Source: IEG data.

The benefits fromirrigation are skewedtoward the better off.

The impact of irrigationon poverty dependsheavily on where thepoverty line is drawn.

has a dynamic effect that is not possible to demon-

strate empirically with IEG’s data. Households sub-

ject to repeated negative shocks become heavily

indebted and deplete their assets, constraining

their ability to undertake productive investments.

Reducing the impact of bad years thus aids asset

accumulation and helps households grow out of

poverty.

The overall impact on intravillage income distri-

bution appears modest and may even be positive,

but the intervillage impact may be less beneficial.

At the simplest level, villages with irrigation grow

faster than those without it, thus creating spatial

inequality. This tendency is reinforced by the fact

that villages in head reaches take excess water,

thus depriving those in the tail end, which are

usually worse off to begin with.27 This tendency

toward spatial inequality is offset (1) to the extent

that those in poorer villages perform labor in

irrigated villages, though this seems limited;

(2) from lower food prices if markets are suffi-

ciently integrated; and (3) by other multiplier

effects (a frequently claimed but little quantified

benefit from irrigation).

Revisiting the ERRA number of factors served to undermine the

economic viability of AP III. First, it was saddled

with the sunk costs of AP II, which need to be

taken into account in assessing the return to the

whole investment in the irrigation scheme. Sec-

ond, AP III was affected by the same cost overruns

and construction delays that had affected AP II.

In addition to these problems, the assumptions

made in the SAR regarding yields and diversifica-

tion were rather optimistic. Combining all these

factors means that the ERR on the money that the

government of Andhra Pradesh and the Bank

have spent in SRSP and SRBC has been consid-

erably less than was expected at appraisal. The re-

vised economic analysis for this study finds an ERR

of just 2 percent, compared with the 19 percent

expected at appraisal (table 4.2).28

4 0


Figure 4.7: Poverty Impact of Irrigation

Pove

rty

mea

sure

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

Upper poverty line Lower poverty line Upper poverty line Lower poverty line

Poverty headcount Poverty gap

Base case (no irrigation)

With project (irrigation)

Source: IEG data.

A Bank study of irrigation in India (World Bank

2003) points out that there are other nonagri-

cultural benefits to irrigation, such as industrial

uses and domestic purposes (drinking water).

Data are not available to quantify these benefits,

but the study shows that for SRSP, 88 percent of

the canal water is used for irrigation purposes.

Therefore, an approximation including the non-

agricultural benefits inflates the benefit stream by

13 percent.29 This raises the overall ERR from 2 to

3 percent (and from 6 to 7 percent excluding AP

II costs).

The yield estimates used in this study are derived

from a number of sources, with remarkable

consistency between those sources. IEG’s final-

estimates of the yield impact are higher than

those used by the

Bank’s irrigation spe-

cialists in preparing

their final review

of AP III. Nonethe-

less, it might be ar-

gued that, with time,

irrigated farmers will

achieve better yields.

Additional calcula-

tions (table 4.3) show

that even if yields with irrigation are as-

sumed to be 20 percent higher than

the increase already estimated, the

overall ERR is still only 4 percent. It re-

mains lower than 10 percent even once

AP II costs are excluded and additional

yields are combined with the nonagricultural

benefits. Even considering the investments from

AP III alone (that is, ignoring AP II costs as sunk

costs), the ERR does not reach the 10 percent

threshold.

The ERR for SRBC is negative, reaching only

zero even when the costs of AP II are excluded,

and 3 percent under the best possible scenario.

All estimates of the ERR for the two components

(that is, those in the SAR and ICR) have shown


4 1

Figure 4.8: Income Distribution Improved between 2005 and 2006

(a) Income Lorenz curves, 2005 and 2006 (b) Lorenz curve for 2005 and concentration curve for 2006

–20

0

0 10 20 30 40 50 60 70 80 90 100

20

40

60

80

100

Cumulative share of population

Cum

ulat

ive

shar

e of

inco

me

20052006

45°

–20

0

0 10 20 30 40 50 60 70 80 90 100

20

40

60

80

100

Cumulative share of population (sorted by 2005 income)Cu

mul

ativ

e sh

are

of in

com

e

20052006

45°

Source: IEG survey.

The revised analysis findsan ERR of only 2 percent,compared with the 19 percent expected at appraisal.

SAR IEG

Excluding ExcludingAll costs AP II ICR All costs AP II

SRBC 9 14 12 –2 0SRSP 31 34 16 5 11Whole project 19 24 15 2 6Source: World Bank data and IEG calculations.

Table 4.2: Estimates of the Economic Rate of Return

this same pattern—of SRBC having a

lower return than SRSP, despite SRSP

serving tail-end users.

Several reasons explain this difference.

First, SRBC was new construction,

whereas SRSP was rehabilitation (the SRSP new

construction component of AP II being dropped

under AP III), and rehabilitation is expected to

yield higher returns than construction. In addi-

tion, SRBC was expensive new construction even

before the cost overruns, because the scheme

will irrigate a relatively small area for the length

of the main canal.30 This difference in perform-

ance between the two components has been

exacerbated by the greater delay surrounding the

SRBC component.

These delays in realizing any benefits in the SRBC

command area mean that any realistic set of as-

sumptions about benefits will not yield a satis-

factory rate of return (unlike SRSP, which nudges

toward an acceptable return as the assumptions

become more generous). Saddled with the cost

overruns from AP II, the aggregate investment in

SRBC never had a chance of achieving economic

viability (figure 4.9).

A major implication of this analysis is

the need to avoid delays and cost over-

runs in major construction projects:

the government of Andhra Pradesh is

taking such steps through the Jalay-

agnam program (see appendix G, at-

tachment 2), which seeks to increase the effi-

ciency of government-financed construction.

A combination of problems rendered the project

unviable, yet the appraisal considered the possi-

bility of cost overruns, delayed benefits, limited di-

versification, and a shortfall in yield gains.

Nonetheless, the appraisal concluded that the

project was robust enough to risk. This is because

it adopted the standard approach to sensitivity

analysis employed in Bank appraisals—of varying

one assumption at a time. The possibility that bad

things may be combined was not considered.31

Using the SAR’s own figures, if all the adverse

shocks considered did occur—which they did,

and more severely than assumed in the SAR’s sen-

sitivity analysis—then the ERR for the whole proj-

ect falls from 24 percent to just 3 percent.32

The sorts of problems encountered are far from

unusual in irrigation projects. The simple step of

calculating the ERR for a situation that combined

all the risks would have cast serious doubt on

the viability of the investment. This was not done,

as Bank procedures are far removed from best

practice (box 4.1).

The problem was exacerbated by some of the

unrealistic assumptions made at appraisal, a prob-

lem highlighted in IEG’s recent irrigation sector

review (IEG 2006b). Ex post returns were gener-

ally found to be lower than those at appraisal be-

cause of overestimation of cropped area, expected

yields, and output prices (IEG 2002).

4 2


Table 4.3: Alternative Estimates of the Economic Rate of Return

SRSP SRBC Whole project

Excluding Excluding ExcludingAll costs AP II All costs AP II All costs AP II

Base case 5 11 –2 0 2 6

Phased yield increasea 6 13 0 2 4 8

Higher yields 6 14 0 2 4 8

Higher yields + nonagricultural benefits 7 16 0 3 5 9Source: World Bank data and IEG calculations.a. Base yield increases in first two years of irrigation, 10 percent higher in years 3–4, and 20 percent higher in years 5 and on.

The appraisal estimateunderestimated risk

because the Bank’sstandard sensitivity

analysis is too simplistic.

The analysis illustrates theirrigation paradox: high

private rates of returncontributing to povertyreduction, but low ERR.

Resolving the Irrigation ParadoxThe irrigation paradox is well illustrated by this

study. The investments in the two schemes sup-

ported by AP II and AP III have yielded a low ERR.

At the same time, the investments have yielded

very high private rates of return to beneficiary

households. The increments in net farm income

and employment have indeed contributed to

poverty reduction. The smoothing of income

made possible by irrigation has helped prevent

households from moving in and out of poverty be-

cause of the vagaries of weather.

The paradox is also readily understood. Farmers

reap the majority of the benefits but bear very few

of the costs. In effect, then, the projects are an in-

come transfer from state coffers (including World

Bank grant and loan funds) to farmers, supple-

mented by subsidies to electricity supply, extension

workers, and so on. Given that the overall finan-

cial and economic returns are negligible, shifting

all costs to farmers would remove the poverty-

reduction impact of irrigation; indeed, poverty

would be increased. The paradox can only be re-

solved by transforming unviable investments into

viable ones. A number of steps will help achieve this.

First, appraisal methods have to undertake a more

rigorous assessment of risks to screen out po-

tentially unviable investments. Second, the plan-

ning about construction time and costs, including

problems in land acquisition, needs to be realis-

tic, and the planning needs to take steps to tackle

the problems that plague these areas. Third, the

assessment needs to be realistic about yield and

diversification benefits. If it is believed that these

really can be achieved with good quality extension

services, then coordination is required (perhaps

through WUAs) to ensure that farmers receive

the information.

If these steps can be undertaken, then invest-

ments may be identified that yield a suitable ERR

and that will yield sufficient income to farmers so

they can meet O&M costs. But

the subsidy element of the cap-

ital cost will remain. So the ques-

tion needs to be asked: Is this

the best use of resources?


4 3

Figure 4.9: Investments in SRBC Had No Chance ofProviding an Acceptable Rate of Return

2.0

1.0

0.0

–1.0

–2.0

–3.0

–4.02000 2002 2004 2006

Year of first benefits

ERR

(%)

2008 2010 2012 201

Cost = 60% of actual

Cost = 80% of actual

Actual costs

The figure shows the ERR for SRBC varying two factors: (1) the extent of cost overruns is dif-ferent for each of the three curves, and (2) the date of first benefits becomes later, moving fromleft to right. Even under the most optimistic scenario the ERR is still less than 3 percent.

Source: World Bank database.

Box 4.1: Improving the Quality ofSensitivity Analysis in BankAppraisals

World Bank appraisal reports conduct deterministicsensitivity analyses, varying just one assumption at atime. Several changes of increasing complexity couldbe made to this approach to improve the quality ofappraisal. The simplest change would be to assume theworst for each assumption considered in the analysis.The ERR is truly robust if it remains acceptable in theface of such analysis. If not, then a more elaborate ap-proach is needed.

The second change would be to assign probabili-ties to the possible outcomes. Once that is done, an ex-pected rate of return can be calculated in astraightforward manner, if it is assumed that the vari-ous risks are independent of one another. If they arenot independent, then the correct approach and bestpractice for sensitivity analysis is to specify the jointdistributions and use Monte Carlo simulations to plotthe distribution of the ERR.

Sources: White 2004; Belli and others 2002.

The paradox can only beresolved by transformingunviable investments intoviable ones.

Chapter 5

Building a brick main canal in India. Photo by Curt Carnemark, courtesy of World Bank Photo Library.

4 7

Summary and Lessons Learned

Two points need be borne in mind when re-

viewing the findings of this study and drawing

lessons from them. First, the projects were “old-

style” irrigation projects of the sort the Bank is

not currently financing in India. Despite this, the

government of Andhra Pradesh is keen to see an

expansion of the area covered by major irrigation

schemes and has launched a substantial program

of new construction. The Bank’s new guidelines

for investment lending in India show that some

of the lessons from the experience of projects

such as AP II and AP III have been learned. Sec-

ond, the reform process establishing WUAs and

defining their roles in practice is likely to be a

long-term one.

Findings on the Six Evaluation Questions

Did investments in irrigated agricultureincrease production in Andhra Pradesh?Yes. Access to irrigation increases both yields and

cropping intensity. Cropping intensity is increased

through the extension of growing into additional

seasons. It also reduces interyear income fluctu-

ations caused by variations in rainfall.

However, the expected diversification into higher-

value crops has not occurred. In fact, irrigated

farmers are less diversified, as they grow more

paddy than farmers whose land is not irrigated,

resulting in excessive water use in the head

reaches.

What are the impacts of PIM on incomes andpoverty alleviation, and who benefits?Gaining access to irrigation raises net farm in-

come by about 60 percent. Indirect benefits come

as well, from additional wage employment. These

indirect benefits are about one-quarter of the

value of the direct benefits.

The top wealth quartile captures the largest por-

tion of the direct benefits and so enjoys the

largest absolute increase in income. But addi-

tional employment income is mostly captured by

the second quartile, which experiences the largest

percentage income growth. The poorest quartile

gains the least absolute benefit but still enjoys in-

come growth of about 20 percent.

Although the largest absolute gain goes to those

who are better off, irrigation results in little change

in the Gini coefficient. The poverty-reduction im-

pact is sensitive to the poverty line. Using an

upper poverty line, irrigation induces a one-off 25

percent reduction in the poverty headcount.

There will be a larger dynamic impact, as irriga-

tion helps smooth productivity and income over

time, breaking the circle of negative income and

debt that can undermine asset accumulation.

Do investments in irrigation yield a satisfactoryeconomic return? In principle, yes; in this case, no. The ex post

rate of return is only 2 percent. The low level re-

sults from cost overruns, construction delays,

and discrepancies between realized income in-

creases and those expected at appraisal.

Two factors are not allowed for in this analysis, one

positive, one negative. First, the ERR calculation

assumes a steady benefit stream. In fact, irrigation

smoothes production income compared with the

“without project” scenario. But the benefits are

only sustained if the canal system is maintained.

Large-scale irrigation will only be economical if

construction costs can be constrained and benefits

sustained. The Bank’s new guidelines prioritize in-

creasing the efficiency of irrigation departments,

which this study confirms to be a necessary start-

ing point for profitable investments.

Was the Bank’s support for canal irrigationrelevant and appropriate?In principle, yes. In practice, however, the an-

swer is twofold: no for the SRBC component and

yes for rehabilitation of SRSP. Irrigation has proven

potential to increase agricultural production and

reduce poverty. Andhra Pradesh has unrealized ir-

rigation potential, so helping meet this potential

appears entirely consistent with a poverty-

reduction strategy—provided, of course, the proj-

ect delivers an acceptable ERR. SRBC failed

to do this, but other aspects of design were in-

appropriate as well.

Excessive faith has been placed in WUAs as a

means of solving problems of both equity in water

distribution and sustainability. But WUAs lack

mechanisms to enforce equitable distribution,

even if they should wish to do so. The physical de-

sign of the system lacked the means for moni-

toring and enforcing proper water usage (that is,

software reform cannot work if the hardware is

missing), and alternative pricing schemes have not

been explored. Increasing the role of WUAs may

improve sustainability, though the evidence shows

that provision for financing O&M matters more

than which bodies are responsible for the works.

It remains possible, of course, that other invest-

ments may have been even more relevant and

appropriate.

Do WUAs facilitate greater participation of the poor and of disadvantaged groups in watermanagement and access to irrigation?No. The poorest are excluded from WUAs be-

cause they are mostly landless. About half of cul-

tivators in canal and tank irrigation schemes report

that they are members of their local WUA. The bet-

ter off are more likely to be members, partly as ir-

rigated land is more likely to belong to the better

off than land in general. Higher castes dominate

the leadership of most WUAs.

Among members, the better off are more likely to

feel they have influence over water management.

WUAs and distributary committees have not pro-

vided an effective mechanism for resolving in-

tercommunity disputes, so the pattern of excessive

water use by farmers in the head reaches con-

tinues, depriving tail-enders, who are usually

poorer, of water. This might have been addressed

by the project committees, but the irrigation bu-

reaucracy has blocked their creation.

Are WUAs an effective means of providingsustainable O&M?It is too early to say. The positive experience with

rehabilitation to date is due to the availability of

external funds (from the Bank’s APERP) to fi-

nance these works. Some of the signs for future

sustainability are not promising: nonpayment or

underpayment of water fees is widespread; most

WUA members do not feel that O&M is their re-

sponsibility; and the amount of money available

to WUAs for O&M is too small, even if they were

to receive their full allocation.

On Selecting Irrigation InvestmentsThe first lessons regard selection of irrigation in-

vestment. More judicious selection would avoid

those projects most likely to have negative returns.

Many previous “lessons learned” still re-main to be learned. The gap between ERR at

appraisal and that found in this study is large.

This has been a consistent finding in IEG reviews

of irrigation projects. The reasons given here are

the same as the reasons found before: under-

estimation of costs and overestimation of bene-

fits. Specifically—

• Construction delays undermine the eco-nomic viability of investments. Serious

delays that postpone the realization of bene-

fits, such as those experienced during these

projects, reduce the rate of return to the in-

vestment. For large-scale irrigation projects to

appear economically attractive, they must be

able to be constructed on time or in a manner

that allows benefits to be phased in at an ear-

lier stage. The government of Andhra Pradesh

4 8


is taking steps to remedy these problems in the

latest phase of irrigation expansion.

• Unrealistic appraisal estimates increasethe likelihood of financing poor invest-ments. The SAR predicted gains from increased

yields and crop diversification that are far from

those that have been realized; the SAR made

the investment appear far more attractive than

it actually was. The Bank’s new guidelines for

irrigation seek to strengthen appraisal proce-

dures, which should directly address this issue.

• The method of risk assessment used inappraisals understates the actual riskthat the project may not be viable. Ap-

praisal reports consider risks one at a time,

not considering the possibility that more than

one of the assumptions in the analysis may

not be realized. Risk assessment methods that

look more rigorously at possible states of the

world will result in more realistic assessments

of project viability.

• Analyses need to value all benefits andcosts. Existing studies (including this one)

neglect some important benefits. The main

ones are increased groundwater levels and

smoothing consumption. All costs should be

valued as well, including realistic full resettle-

ment costs, adverse environmental impacts

(which are most likely long term), and nega-

tive effects on those deprived of water by irri-

gation elsewhere. The Bank’s new guidelines

explicitly recognize the importance of in-

cluding all costs and benefits in the appraisal

analysis.

On Participatory Irrigation ManagementThe lessons regarding PIM are generally applica-

ble to most forms of participatory (community-

based) development.

• Bottom-up development needs supportfrom the top. The initial burst of WUA activ-

ity reflected strong support from the state gov-

ernment. Once the government changed, there

was less support for the program, so it lacked

the central direction needed to bring about

change in the bureaucracy and to support

WUAs as a pro-poor institution at the grassroots

level.

• PIM may alleviate the problems of large-scale irrigation schemes, but it cannoteliminate them. Several of the problems

encountered in these projects, such as con-

struction delays and weakness of extension

services, will not be solved by the introduc-

tion of WUAs. Moreover, WUAs are not real-

izing a potential benefit, as funds for O&M are

insufficient.

• PIM is not intrinsically pro-poor. Farmers

benefiting from irrigation are among the less

poor, and the better-off farmers are most likely

to participate in WUAs. Hence, WUAs are more

likely to reinforce existing social patterns than

to change them.

• WUAs have limited means to effectivelyresolve water allocation disputes.

On the Impact of IrrigationFinally, there are lessons on the design of irriga-

tion projects.

• Irrigation raises output through higheryields and cropping intensity and stabi-lizes production, all of which contributeto poverty reduction. But irrigation in-vestments constitute a sizeable subsidy tofarmers who are not among the poorestor even the poor.

• The direct income benefits to farmersfrom irrigation provide the potentialfor full recovery of the recurrent costs of the services provided. Current water

charges are a tiny fraction of the net increment

in income that irrigation makes possible. In-

creasing these charges is necessary for sus-

tainability of the system.

• Incentive structures, including waterpricing policy, encourage water wastage.

S U M M A R Y A N D L E S S O N S L E A R N E D

4 9

Appendixes

Workers in a rice field. Photo by Thomas Sennett, courtesy of World Bank Photo Library.

5 3

The Independent Evaluation Group (IEG) study

design required the sample to be split into three

groups: villages receiving canal irrigation since

before the first-round survey in June 2005 (Group

1); villages receiving canal irrigation for the first

time in the two agricultural seasons prior to the

second-round survey, that is, between rounds one

and two of the survey (Group 2); and villages not

receiving canal irrigation yet, but scheduled to re-

ceive it in the near future (Group 3) (table A.1).

The rationale for this design is as follows. Com-

paring the treatment group (Group 2) with the

second control group (untreated; Group 3) will

allow for a standard double-difference estimation

of project effect. The control was selected from

villages scheduled to receive irrigation in line

with a “pipeline approach,” that is, the assump-

tion that those selected for participation but not

yet treated are more likely to share characteristics

with the treatment group than would a randomly

selected control sample. However, there is a con-

cern that farmers take time to adapt their farm-

ing to the availability of irrigation. Control 1

(Group 1) provides a group that has received at

least two treatments (years of irrigation), allow-

ing single-difference estimates with the untreated

group.

AP II and AP III financed work in two command

areas, Srisailam Right Branch Canal (SRBC) and the

Sriramasager Project (SRSP). The study originally

intended to select the sample from an area served

by the new SRBC canal. However, at the time of

the first round of the survey, a year after Andhra

Pradesh III (AP III) had closed, very little water had

been released (irrigating 7,800 hectares [ha] of a

planned 65,000 ha) in the SRBC command, be-

cause the micro-network was at best 20 percent

complete. Completion of the network was esti-

mated for 2008 (now expected for 2007), so the

prospects of a much larger irrigated area by the

time of the 2006 second round were bleak. The

sample was therefore restricted to the SRSP com-

mand, where the main canal and several distrib-

utaries are being extended.

Specifically, the study focused on areas irrigated

by SRSP between kilometer (km) 234 and km

284. This area falls entirely within Warangal district.

Farmers on these distributaries are at various

stages of receiving water: Some had received

water by the time of the first round; some were

to get it by the second round, and some at some

point later. These characteristics fit with the de-

sired three-group sample design. Hence, the sam-

ple design consisted of three groups of six villages

each: (a) Control 1—water released for the first

time in kharif 2003, (b) Project (treatment group):

water released in kharif 2005, and (c) Control 2:

no water released yet.

In this range, other than distributary canal 31 (DC

31), the new canal construction work is not funded

by the Bank. Moreover, DC 31 is not suitable for

inclusion in the sample, as all works were com-

pleted prior to kharif 2003; that is, all villages on

the canal are in Group 1. However, Bank-funded

rehabilitation of the main SRSP canal is what has

made water releases south of km 234 possible

(see chapter 2). Hence, the sample selection is

APPENDIX A: SAMPLE SELECTION OF IEG IRRIGATION SURVEY

Table A.1: Scheme for Survey Coverage

Round 1 Round 2

Group 1 (Control 1)

Group 2 (treatment)

Group 3 (Control 2)Note: Shading indicates that the group has been exposed to irrigation.

justified on three grounds. First, reaching the

areas served has been made possible by Bank-

funded activities, even if the Bank has not directly

financed the canal extensions. Second, the im-

pact of irrigation in the areas may reasonably be

expected to be similar to those in areas which

were financed by the Bank. Third, a major focus

of this study is the economic justification for in-

vesting in further new irrigation works in the state.

Because access to irrigation is determined by

topology, this analogy to a pipeline approach ap-

pears promising. However, it is known that access

to water differs between the head and tail reaches,1

and evidence from the pilot confirmed the preva-

lence of water theft/inefficient usage in upper

reaches. Unfortunately for the evaluation design,

the coverage of the canal system is being ex-

tended mainly by developing the tail reaches of

the longer distributaries (the longest being DC 38

at 80 km in length). Hence, as shown in table

A.2, potential treatment and control sample vil-

lages are not evenly spread across the various

reaches. Control 1 villages are concentrated in the

head reaches and Control 2 villages in the tails.

The exceptions come because there are some

short canals (DC 40 and 41 at around 10 km in

length each), which can be counted as “all head.”

Moreover, other characteristics, which are very

likely correlated with project outcomes, also vary

systematically between project areas and the po-

tential controls. Broadly speaking, Control 1 vil-

lages are most developed and Control 2 least so,

with the treatment group falling in the middle;

see the various indicators in table A.3. Control 2

also has a much higher share of scheduled tribes,

some villages being close to 100 percent sched-

uled tribe communities.

The initial sample design was to stratify by reach

and select even numbers for each group from

head and tail (leaving out the middle reach),

although the sample size was small for Con-

trol 2, especially for head reaches.2 However,

this procedure did not eliminate the systematic

differences between the three groups, so the

validity of the comparison could be called into

question.

Hence, a matching procedure based on propen-

sity score matching was adopted. A probit re-

gression was estimated using census data of “the

propensity to be irrigated”; that is, villages in the

treatment group were given a value of one for the

dependent variable. Irrigation status for each vil-

lage was taken from a list provided by the Irriga-

tion Department in Warangal district. The

regressors were the variables shown in table A.3.

A population-weighted random sample of 12 proj-

ect villages was then drawn, and the nearest neigh-

bor village (based on propensity score) from the

other two groups identified. The 12 were reduced

5 4


Table A.2: Distribution of Villages by Reach

Head Middle Tail Unclassified Total

Number

Control 1 55 29 5 4 93

Project 26 71 30 5 132

Control 2 3 1 15 0 19

Percent

Control 1 59 31 5 4 100

Project 20 54 23 4 100

Control 2 16 5 79 0 100Source: Calculated from Andhra Pradesh census data, 2000.

Table A.3: Socioeconomic Characteristics of Projectand Potential Control Villages (group averages)

Control 1 Project Control 2

Literacy (%) 48.3 45.4 39.1

Male literacy (%) 61.0 57.5 48.9

Female literacy (%) 35.0 33.2 29.0

No. of educational institutions 1.00 1.00 0.94

No. of medical institutions 1.10 1.08 0.94

Drinking water 1.00 1.00 0.94

Pucca road 0.74 0.65 0.88

Distance to Warangal (km) 37.1 60.1 84.3

Percent scheduled castes 14.2 15.9 10.9

Percent scheduled tribes 19.1 21.6 35.8

Percent of land cultivated 21.8 19.4 16.0

Percent agriculture laborers 24.7 26.9 30.5

Sex ratio (women/1,000 men) 961 974 912Source: Calculated from Andhra Pradesh census 2000.

to 8 to avoid duplicate matches from Control 2

(but even so only 7 villages were picked from

this group). Inspection of the actual values of the

socioeconomic characteristics was used to re-

move the less typical villages, resulting in a sam-

ple of 18, 6 in each group. This procedure

considerably reduced intergroup variation in

socioeconomic characteristics, and where it re-

mained, it did not follow the systematic pattern

shown in table A.3.

Unfortunately, the sample selected was prob-

lematic; field visits revealed that Irrigation De-

partment data on irrigation status of each village

were not reliable. To rectify this situation, a list was

made of three to four possible matches for each

of the two controls, again using the closest match

based on the propensity scores. The actual situ-

ation in these villages was verified in the field, and

thus the final sample selected. But even then vil-

lages that were due to receive water between

the surveys (that is, to be in the project group)

turned out not to receive water.

There are two possible ways to handle this situa-

tion. The first is to include the tail-end villages in

the treatment group, which would give an im-

pact measure of the “intention to treat.” The sec-

ond approach is to recategorize the tail enders into

the “no water” group (Control 1), so the impact

estimate is that for “treatment of the treated.”

But since tail-end villages are unlike other vil-

lages, this latter approach reintroduces systematic

differences into the pre-existing characteristics

of the sample villages, as clearly shown in table A.4.

However (as argued at greater length in appen-

dix C), we know the direction of bias. Tail-end

villages have less access to markets, less well-

educated populations, and are less visited by

extension workers, so we would expect the im-

pact of irrigation to be less than in the head

reaches. Hence, using impact estimates based on

the head reaches produces an upward bias in

those estimates. In addition, the treatment of the

treated impact estimate of course exceeds that

from intention to treat. Nonetheless, it is the

treatment of the treated estimate for the higher

reaches that is used here, because—despite the

upward bias in impact (that is, we have an upper

bound estimate)—the rate of return to the proj-

ect is low.

A P P E N D I X A : S A M P L E S E L E C T I O N O F I E G I R R I G AT I O N S U R V E Y

5 5

Table A.4: Characteristics of Selected Villages

Group

No Newirrigation irrigation Irrigation

Literacy (%) 39 47 42

Male literacy (%) 50 58 54

Female literacy (%) 28 35 31

No. of educational institutions 1.00 1.00 1.00

No. of medical institutions 1.00 1.00 1.00

Drinking water 1.00 1.00 1.00

Pucca road 0.89 0.17 0.17

Distance to Warangal (km) 71 36 44

Percent scheduled castes 12 19 15

Percent scheduled tribes 37 15 29

Percent of land cultivated 19 17 25

Percent agriculture laborers 30 29 24

Sex ratio (women/men) 967 968 977Source: Calculated from Andhra Pradesh census data, 2000.

5 7

This chapter reports the full regressions for the results discussed in chapters 3 and 4. Estimation pro-

cedures are listed in the title of each table. All data for these regressions are from the IEG survey.

APPENDIX B: REGRESSION RESULTS

Table B.1: Determinants of WUA Membership, 2006 (probit regression)

Cultivating households only All households

Coefficient Coefficient(t-stat) (t-stat)

Caste (ref = scheduled caste)

Scheduled tribes –0.60** –0.33

(–2.17) (–1.31)

Other backward caste –0.43** –0.31

(–1.79) (–1.49)

Other caste –0.70** –0.56*

(–2.02) (–1.70)

Wealth (ref = 1st, poorest quartile)

2 0.98** 1.02***

(2.57) (3.12)

3 1.16** 1.23***

(2.95) (3.70)

4 1.95*** 1.96***

(5.01) (5.70)

Land (acres) 0.01 0.04

(0.21) (1.22)

Education (ref = none)

Primary 0.25 0.18

(1.25) (1.06)

Secondary or higher 0.35 0.21

(1.42) (0.91)

Intercept –1.62*** –1.90

(–3.94) (–5.49)Notes: The regression includes village dummies that are not shown. Estimation drops the three villages in which there are no WUA members. These village dummies are the identifying variables for the selection equation for the two stage estimates reported in tableB.2; t statistics in parentheses.

*Significant at 10 percent.**Significant at 5 percent.

***Significant at 1 percent.

5 8


Table B.2: Determinants of Participation in WUA Activities, 2006 (probit and ordered probit,adjusted for selection into WUA membership)

Regularly Take part FeelAttended attend Taken in WUA influence Voted Voted

WUA WUA part in decision water for WUA for TCmeeting meeting O&M making management president chair

Wealth quartile (ref category = 1)

2 0.57 0.14 4.71 4.33 4.66 0.21 0.10

–1.36 –0.32 (.) (.) (.) –0.61 –0.29

3 0.321 0.27 4.39 4.02 3.67 0.35 0.34

–0.79 –0.68 (13.06)*** (12.16)*** (7.56)*** –0.96 –0.95

4 0.67 0.55 4.30 3.82 3.97 0.36 0.43

–1.63 –1.26 (14.76)*** (10.07)*** (11.59)*** –0.94 –1.16

Education level (ref category = none)

Primary 0.53 0.69 0.36 0.46 0.18 –0.057 0.02

(2.62)*** (2.70)*** –1.58 (1.78)* –0.62 –0.29 –0.10

Secondary or higher 0.40 0.32 –0.32 –0.29 –0.88 0.08 0.13

–1.47 –0.96 –0.84 –0.79 (1.88)* –0.33 –0.52

Caste (ref category = scheduled caste)

Scheduled tribe –0.52 –0.36 0.65 0.18 –0.21 –0.13 –0.11

–1.55 –0.89 –1.41 –0.29 –0.34 –0.49 –0.45

Other backward caste –0.53 –0.56 0.86 0.16 0.46 –0.13 –0.08

(2.05)** –1.64 (2.35)** –0.34 –1.12 –0.51 –0.34

Other caste –0.69 –0.26 0.83 0.64 0.29 –0.03 0.01

(1.69)* –0.57 –1.64 –1.06 –0.43 –0.08 –0.02

Land (acres) –0.003 0.01 –0.01 0.01 0.01 –0.002 –0.01

–0.14 –0.33 –0.49 –0.27 –0.4 –0.07 –0.18

Lambda –5.343 –3.409 –5.118 –3.975 –3.838 –6.564 –6.104

(5.90)*** (3.63)*** (5.41)*** (3.75)*** (3.87)*** (7.59)*** (7.32)***

Constant 1.787 0.177 –3.414 –3.811 –3.88 3.124 2.733

(2.23)** –0.22 (5.35)*** (4.40)*** (4.95)*** (4.13)*** (3.80)***

Observations 566 566 566 566 566 566 566Note: Sample is WUA members only. Absolute value of t statistics in parentheses. WUA = water user association.

* Significant at 10 percent. ** Significant at 5 percent.

*** Significant at 1 percent.

A P P E N D I X B : R E G R E S S I O N R E S U LT S

5 9

Table B.4: Dependent Variable: Value ofTotal Farm Output

Coefficient t-stat

Log of land (ha) 0.57 13.25***

Log of capital (service value) 0.02 1.89*

Log of draught animals (value) 0.02 2.03**

Log of male days own labor 0.07 2.29**

Log of female days own labor 0.04 3.30***

Log of female days hired labor 0.10 3.10***

Log of male days hired labor 0.05 1.79*

Log of fertilizer (expenditure) 0.09 3.15***

Log of seeds (expenditure) –0.02 –1.07

Log of pesticides (expenditure) –0.01 –0.63

Log of other inputs (expenditure) 0.00 –0.05

Rabi season (dummy) 0.13 2.17**

Summer season (dummy) 0.03 0.34

Permanent crops (dummy) 0.13 1.00

Log of age of head of household 0.06 0.60

Log of education of head of household 0.11 2.53***

Share of canal-irrigated land 0.04 0.25

Share of tank-irrigated land 0.43 3.60***

Share of tubewell-irrigated land 0.30 2.45***

Share of borewell-irrigated land 0.29 2.65***

Crop shock (dummy) –0.19 –2.84***

Log of distance from market (km) –0.01 –0.46

Constant 7.66 17.53***

Observations 506.00

R-square 0.68Note: Value in rupees; ha = hectares.***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.

Table B.3: Determinants of Perception thatWater Is Fairly Distributed within Village(probit using 2006 data)

(1) (2)

Wealth quartile (ref = 1)

Quartile 2 –0.10 –0.11

Quartile 3 0.10 0.01

Quartile 4 0.14 0.01

Education level (ref = none)

Primary 0.07 0.09

Secondary or higher –0.26 –0.22

Land (ref = landless)

Marginal 0.96*** 0.97***

Small 1.09*** 1.14***

Large 1.22*** 1.25***

WUA

Household member 0.76*** 0.80***

Village membership rate –0.48 –1.38**

Irrigation characteristics

Proportion household land irrigated 0.84* . .

Number of waterings of irrigatedplots 0.07*** . .

Village water adequacy . . 0.98

Intercept –2.63*** –2.88***

Number of observations 896 896Note: . . = negligible.***Significant at 10 percent. ***Significant at 5 percent. ***Significant at 1 percent.

6 0


Table B.5: Dependent Variable: Production per Hectare

Coefficient t-stat

Log of land (ha) –0.30 –8.98***

Log of capital (service value) 0.02 2.44**

Log of labor (days) 0.13 3.45***

Log of all inputs (expenditure) 0.15 4.52***

Log of age of head of household –2.22 –1.57

Log of age of head of household squared 0.31 1.63*

Log of education of head of household 0.11 3.64***

Red sandy soil (dummy) 0.11 1.89*

Black soil (dummy) 0.18 3.28***

Chaudu soil (dummy) 0.18 3.12***

Irrigation sources (number) –0.17 –2.66**

Rabi season crop (dummy) –0.32 –5.72***

Summer season crop (dummy) –0.44 –4.77***

Crop shock (dummy) –0.13 –2.94***

Rice crop 1.12 7.55***

Maize crop 0.66 4.24***

Groundnut crop 0.24 1.50

Cotton crop 0.44 3.39***

Chilies crop 0.14 0.57

Turmeric crop 0.69 4.74***

Pulses crop –0.31 –2.09**

Canal-irrigated field (dummy) –0.50 –2.44**

Tank-irrigated field (dummy) 0.29 2.92***

Groundwater-irrigated field (dummy) 0.00 0.02

Canal x rice 0.64 2.92***

Canal x maize 1.09 2.81**

Canal x groundnut 0.52 1.33

Canal x chilies 0.67 1.90*

Canal x turmeric 1.04 2.31**

Groundwater x rice 0.15 1.71*

Groundwater x chilies 0.45 2.11**

Constant 8.95 3.36***

Observations 1,113.00

R-square 0.45Note: Production measured in kilograms; ha = hectares.***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.


6 1

Table B.6: Dependent Variable: Value of Fertilizer Expenditure

Coefficient t-stat

Land (ha) 94.43 1.39

Canal-irrigated land (ha) 660.11 3.60***

Tank-irrigated land (ha) 692.34 4.65***

Tubewell-irrigated land (ha) 399.42 3.29***

Borewell-irrigated land (ha) 435.32 3.41***

Pumpset (dummy) 180.75 0.43

Rabi season (dummy) –299.47 –1.76*

Summer season (dummy) 355.22 1.39

Red sandy soil (dummy) 186.99 2.66***

Black soil (dummy) 232.92 3.34***

Chaudu soil (dummy) 280.62 3.38***

Extension (dummy) –181.51 –0.21

Female-headed household –1,784.16 –2.11**

Age of head of household 4.49 0.29

Education of head of household 22.17 0.18

Cultivators in household (number) 315.56 1.92*

Laborers in household (number) –345.65 –2.39**

Off-farm nonagricultural workers (number) –722.60 –2.29**

Scheduled tribe –1,550.92 –1.84*

Scheduled caste –1,600.94 –2.16**

Other backward caste –1,037.31 –1.51

Wealth index 2457.82 1.32

Draught animals (number) 144.72 0.81

Sheep (number) 47.32 0.44

Goats (number) 71.92 1.07

Distance to fertilizer shop (km) –33.19 –1.11

Share of maize land –1,950.32 –1.40

Share of groundnut land 1,828.03 1.33

Share of cotton land 617.52 1.05

Share of chilies land 3,526.50 2.90***

Share of turmeric land 3,569.64 2.86***

Share of pulses land –83.87 –0.08

Share of other crops 1,616.61 0.79

Male wage (village) –19.73 –0.22

Constant (village) 6,841.99 1.07

Observations 491.00

R-square 0.56Note: Village level price variables for major crops were included but not shown as insignificant; ha = hectares. ***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.

6 2


Table B.7: Dependent Variable: Value All Inputs Expenditure

Coefficient t-stat

Land (ha) 43.29 0.30Canal-irrigated land (ha) 2,080.30 5.63***Tank-irrigated land (ha) 1,833.91 6.24***Tubewell-irrigated land (ha) 2,086.32 8.64***Borewell-irrigated land (ha) 1,674.16 6.71***Pumpset (dummy) –74.25 –0.08Rabi season (dummy) –1463.38 –4.46***Summer season (dummy) –485.95 –0.89Red sandy soil (dummy) 422.87 2.84***Black soil (dummy) 530.38 3.57***Chaudu soil (dummy) 848.73 4.78***Extension (dummy) 2,163.85 1.17Female-headed household –3,443.59 –1.95**Age of head of household 24.19 0.73Education of head of household –94.22 –0.35Cultivators in household (number) –86.85 –0.25Laborers in household (number) –302.06 –0.98Off-farm nonagricultural workers (number) –1,437.84 –2.20**Scheduled tribe 326.18 0.19Scheduled caste –305.26 –0.20Other backward caste 1,166.22 0.82Wealth index 8,714.84 2.21**Draught animals (number) –267.84 –0.70Sheep (number) 722.09 3.07***Goats (number) 10.67 0.07Distance to fertilizer shop (km) –44.89 –0.71Share of maize land –271.01 –0.09Share of groundnut land 4,623.78 1.55Share of cotton land 2,468.32 1.99**Share of chilies land 9,236.46 3.52***Share of turmeric land 5,975.32 2.20**Share of pulses land 801.12 0.36Share of other crops 6,347.78 1.44Rice price (village) –5.14 –0.70Maize price (village) –4.58 –1.28Groundnut price (village) 0.71 0.40Cotton price (village) 6.57 0.84Chilies price (village) 1.16 1.14Turmeric price (village) –0.46 –0.54Pulses price (village) –1.09 –1.01Other prices (village) 0.40 0.57Male wage (village) 55.17 0.28Constant (village) –1,0748.02 –0.83Observations 514R-square 0.60Note: Value in rupees.***Significant at 1 percent.**Significant at 5 percent.


6 3

Table B.8: Total Household Labor Demand (days)

Coefficient t-stat

Land cropped –0.40 –0.06

Canal-irrigated land 58.10 3.42***

Tank-irrigated land 51.02 6.32***

Tubewell-irrigated land 86.34 8.40***

Borewell-irrigated land 40.07 3.63***

Share of maize land –10.93 –0.07

Share of groundnut land 21.12 0.13


Share of chilies land 120.60 0.92

Share of turmeric land –12.27 –0.09


Share of other crops 557.04 2.36**

Share of red sandy soil –172.12 –2.63***

Share of black soil –12.59 –0.22

Share of chaudu soil –85.87 –1.49

Draught animals (number) 48.78 2.56***

Goats (number) –5.96 –0.79

Education of head of household –16.52 –1.14

Female-headed household –137.53 –1.35


Female children (number) –14.37 –0.61

Male adults (number) 11.35 0.40

Female adults (number) 101.85 2.92***

Elderly male (number) –35.25 –0.51

Elderly female (number) –44.65 –0.68

Scheduled tribe 105.56 1.16

Scheduled caste 66.65 0.83

Other backward caste 84.02 1.18

Wealth index 412.22 2.14**


Constant 3.50 0.01

Observations 471

R-square 0.52***Significant at 1 percent.**Significant at 5 percent.

6 4


Table B.9: Dependent Variable: Demand for Own HouseholdLabor (days)

Coefficient t-stat

Land cropped 5.66 1.82*

Canal-irrigated land 5.24 0.66

Tank-irrigated land –4.44 –1.17

Tubewell-irrigated land 1.37 0.29

Borewell-irrigated land –0.78 –0.15

Share of maize land 43.28 0.59

Share of groundnut land 111.81 1.44


Share of chilies land 102.92 1.67*

Share of turmeric land 35.54 0.56


Share of other crops –14.72 –0.13

Share of red sandy soil –148.50 –4.82***

Share of black soil –20.49 –0.75

Share of chaudu soil –95.51 –3.51***

Draught animals (number) 27.37 3.06***



Female headed household –17.35 –0.35


Female children (number) 1.11 0.10

Male adults (number) 33.60 2.52**

Female adults (number) 87.59 5.33***

Elderly male (number) 69.30 2.11**


Scheduled tribe 86.04 2.00**

Scheduled caste 80.59 2.12**

Other backward caste 93.99 2.82**

Wealth index 61.59 0.68


Constant 438.61 2.05

Observations 464

R-square 0.33***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.


6 5

Table B.10: Dependent Variable: Hired Labor Demand (days) (Tobit)

Coefficient t-stat

Land cropped –6.29 –0.95

Canal-irrigated land 52.67 3.15***

Tank-irrigated land 55.44 6.90***

Tubewell-irrigated land 85.03 8.35***

Borewell-irrigated land 41.15 3.73***

Share of maize land –71.53 –0.41

Share of groundnut land –92.73 –0.56


Share of chilies land 8.25 0.06

Share of turmeric land –58.17 –0.43

Share of pulses land 32.24 0.20

Share of other crops 569.70 2.44**

Share of red sandy soil –39.39 –0.59

Share of black soil 4.09 0.07

Share of chaudu soil 5.08 0.09

Draught animals (number) 21.64 1.12



Female headed household –112.38 –1.12


Female children (number) –12.63 –0.53

Male adults (number) –21.82 –0.76

Female adults (number) 18.75 0.51

Elderly male (number) –101.37 –1.45



Scheduled caste –14.08 –0.17

Other backward caste –10.16 –0.14

Wealth index 358.65 1.83*

Male wage (village) 10.31 0.80

Constant –521.68 –1.12

Observations 435

R-square 0.48***Significant at 1 percent.**Significant at 5 percent.*Significant at 10 percent.

6 7

Single-Difference Estimates fromSecondary SourcesSingle-difference estimates of the impact of irri-

gation on crop production are available by com-

paring irrigated and rain-fed plots in comparable

areas. Geographic proximity is taken to ensure

comparability. Such estimates are the most com-

mon method of estimating the impact of irriga-

tion. Single difference estimates are reported

here from the economic analysis for the AP III Im-

plementation Completion Report (ICR), IEG’s

own reestimation following the ICR methodology,

the AP III Baseline Survey, and the IEG survey data.

The SRSP command covers three districts: Karim-

nagar, Nizamabad, and Warangal. To estimate pro-

duction impacts in SRSP, the ICR for AP III used data

for Karimnagar district, comparing data from the 16

mandals that fall entirely within the SRSP com-

mand with the 22 mandals located outside the

SRSP command area and that do not have any sig-

nificant surface irrigation sources. The SRBC com-

mand falls within Kurnool district, from which data

were used for the five mandals accounting for 90

percent of the command area. Because no irriga-

tion water had been received at the time this report

was prepared, these data are the “without project”

scenario. The “with project” data are assessed based

on ICR mission’s field visits and discussions. The

SRBC estimates used in the ICR are therefore spec-

ulative rather than being based on firm data.

Table C.1 summarizes the results from the ICR’s

analysis regarding yields, and table C.2 those for

cropping patterns and intensity. The years used

for the analysis are not stated.

Tables C.3 and C.4 present IEG’s own calcula-

tions of the same analysis on an annual basis for

the years 2002–03 to 2004–05 regarding crop-

ping patterns and intensity. The data show the fall-

off in production in the last year, the last year of

the drought in which no water was released in the

SRSP command. This year can be seen as a natu-

ral experiment, the fall in yields (table C.5) partly

capturing an “after versus before” irrigation effect

(but also showing the impact of general water

shortage in unirrigated areas).

The IEG results differ somewhat from those in the

ICR. The command area cropping intensity is

similar, but lower in the noncommand compari-

son group mandals. In the IEG data the absolute

share of paddy is lower in both treatment and

comparison groups and the difference smaller. But

the paddy remains more common in irrigated

areas than unirrigated, especially in the kharif sea-

son (paddy has quite a high share of noncommand

area rabi production because rabi production is

necessarily irrigated). Finally, the reported yields

in the ICR are higher than those in the more re-

cent data, though the “natural experiment” of no

SRSP releases in the last year gives very similar

yield impact estimates to those from the SRSP

baseline, which are reported next.

The second set of single-difference estimates is

from the AP III baseline report for SRSP. The yields

for the command area are based on crop-cutting

experiments carried out by the Agriculture De-

partment, and those for noncommand areas in the

three districts are based on the data from the Di-

rectorate of Economics and Statistics. Where these

estimates are not available, SRSP estimates are

compared with the state-level estimates. Table

C.6 reports the yield estimates obtained in this

manner.

A final set of single difference estimates comes from

the IEG survey data, taking the comparison of

APPENDIX C: SINGLE- AND DOUBLE-DIFFERENCE IMPACT ESTIMATES

(text continues on page 71)

6 8


Table C.1: Single Difference Yield Estimates (kg 000/ha)

SRSP SRBC Percent increase

WOP WP WOP WP SRSP SRBC

Paddy (kharif ) 4.0 4.6 4.5 5.1 15.0 13.3Paddy (rabi ) 4.3 5.1 4.0 5.4 18.6 35.0Maize (kharif ) 3.3 3.9 n.a. n.a. 18.2 n.a.Maize (rabi ) 4.0 5.0 n.a. n.a. 25.0 n.a.Cotton (kharif ) 1.0 1.5 1.0 1.5 50.0 50.0Chilies (kharif ) n.a. 1.5 n.a. 1.5 n.a. n.a.Chilies (rabi ) n.a. 1.5 n.a. 1.5 n.a. n.a.Groundnut (kharif ) n.a. n.a. 1.0 2.0 n.a. 100.0Groundnut (rabi ) 1.1 1.7 1.5 2.5 54.5 66.7Groundnut (kharif ) n.a. n.a. 1.0 2.0 n.a. 100.0Sunflower (kharif ) n.a. n.a. 0.6 1.2 n.a. 100.0Sunflower (rabi ) n.a. n.a. 1.0 2.0 n.a. 100.0Sorghum (kharif ) n.a. n.a. 1.0 2.0 n.a. 100.0Sorghum (rabi ) n.a. n.a. 1.5 2.5 n.a. 66.7Chickpea (rabi ) n.a. n.a. 1.0 2.0 n.a. 100.0Source: Economic and financial analysis working paper for AP III ICR.Note: WOP = without project; WP = with project; n.a. = not applicable.

Table C.2: Single-Difference Cropping Pattern (share ofcropped area) and Intensity

SRSP SRBC

WOP WP WOP WP

Cropping patternsKharif

Paddy 42 70 2 4Sorghum 0 0 5 4Sunflower 0 0 5 15Maize 6 9 0 0Groundnut 0 0 2 22Cotton 12 20 15 5Chilies 0 1 0 5Rain-fed crops 40 0 0 0Total 100 100 29 55

RabiPaddy 21 32 2 1Sunflower 0 0 6 25Sorghum 0 0 25 0Chickpea 0 0 40 35Maize 5 7 0 0Groundnut 5 6 1 29Chilies 0 1 0 5Total 31 46 74 95

Cropping intensity 130 145 103 150Source: Economic and financial analysis working paper for AP III ICR.Note: WOP = without project; WP = with project.

Table C.4: Crop Shares of Cropped Area, by Season

2002–03 2003–04 2004–05

Kharif Rabi Kharif Rabi Kharif Rabi TotalCommand area

Paddy 52 53 46 54 29 21 44

Groundnut 0 11 0 10 0 23 4

Cotton 17 0 19 0 35 0 16

Chilies 3 0 4 0 3 0 2

Turmeric 3 0 3 0 5 0 3

Total 100 100 100 100 100 100 100

Memo items

Total (000 ha) 117 49 119 63 93 31 472

Total as % 2002–03 100 41 102 53 79 27

Noncommand area

Paddy 32 38 17 29 13 25 21

Groundnut 2 16 1 12 1 18 3

Cotton 18 0 14 0 39 0 20

Chilies 3 3 2 4 3 0 3

Turmeric 0 0 0 0 0 0 0

100 100 100 100 100 100 100

Memo items

Total (000 ha) 88 17 117 26 119 17 384

Total as % 2002–03 100 19 133 30 135 20Source: Directorate of Economics and Statistics.Note: ha = hectares.

A P P E N D I X C : S I N G L E - A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S

6 9

Table C.3: Single-Difference Estimate of Cropping Intensity (ratio of gross croppedarea to net area sown)

Table C.5: Karimnagar District Average Yields (kg/ha), 2000–01 through 2004–05

2000–01 2001–02 2002–03 2003–04 2004–05

Kharif Rabi Kharif Rabi Kharif Rabi Kharif Rabi Kharif RabiRice 3.1 3.1 3.0 3.5 2.6 3.2 3.2 3.2 2.3 2.8

Jowar 0.7 1.0 1.0 1.1 1.0 0.8 0.9 0.9 0.9 1.0

Maize 3.2 4.0 3.5 5.0 3.1 4.4 3.9 4.7 2.3 3.5

Groundnut 0.6 1.9 0.6 2.0 0.5 1.6 0.5 2.1 0.9 1.0

Chilies 2.2 2.3 1.9 2.8 1.4 2.5 1.7 3.3 1.4 3.0

Cotton 0.4 0.2 0.3 0.4 0.2Source: Directorate of Economics and Statistics.

2002–03 2003–04 2004–05 Average

Command mandals 141 151 134 142

Noncommand mandals 119 123 116 119

Single difference 22 28 18 23

Kharif RabiNon- Non-

CADA CADA % difference CADA CADA % difference

Paddy

1990–2000 3.1 2.4 27 3.4 2.5 38

1998–99 3.1 2.7 13 3.5 2.9 22

1997–98 2.6 1.9 36 2.9 2.5 18

1996–97 3.0 2.6 14 3.3 2.8 18

1995–96 2.2 2.2 0 2.8 2.6 5

1994–95 2.7 2.5 8 2.9 2.4 21

Maize

1990–2000 3.8 3.3 15 4.2 3.7 13

1998–99 3.7 3.3 11 5.2 3.6 46

1997–98 2.6 1.9 36 3.6 2.7 36

1996–97 3.0 2.6 14 4.8 3.8 26

1995–96 2.2 2.2 0 3.8 3.1 25

1994–95 2.7 2.5 8 3.6 3.2 14

Groundnut

1990–2000 n.a. n.a. n.a. 1.6 1.2 35

1998–99 n.a. n.a. n.a. 1.5 1.5 1

1997–98 n.a. n.a. n.a. 1.1 0.9 17

1996–97 n.a. n.a. n.a. 1.7 1.5 13

1995–96 n.a. n.a. n.a. 1.4 1.0 46

1994–95 n.a. n.a. n.a. 1.6 1.1 43

Single-Difference Yield Estimates, SRSP (kg 000/ha)

Turmeric Chilies Cotton

Percent Percent PercentSRSP State difference SRSP State difference SRSP State difference

1994–95 0.31 0.29 9

1995–96 4.8 4.9 –2 1.6 1.8 –13.0 0.34 0.26 31

1996–97 5.0 4.2 19 2.3 2.2 5.0 0.36 0.32 14

1997–98 5.8 5.7 2 1.6 2.0 –22.0 0.23 0.25 –6

1998–99 5.5 5.6 –2 2.1 2.4 –13.0 0.35 0.20 75

1990–2000 4.2 4.7 –10 1.8 1.8 0 0.34 0.26 30Source: AFC (Hyderabad) AP III baseline survey.Note: CADA = Command Area Development Authority, which includes but is greater than the irrigated area; ha = hectares; kg = kilogram; n.a. = not applicable.

Table C.6: Single-Difference Yield Estimates, SRSP (kg 000/ha)

7 0


yields from irrigated plots and unirrigated ones

using the 2006 data. These data were collected only

in Warangal district. In this presentation (table

C.7), “unirrigated” excludes plots irrigated from

groundwater by either tubewells or boreholes.

The impact is largest for paddy in rabi, given the

very low reported yield in the unirrigated plots.

There are three sources of bias in the single-

difference estimates. First, it is assumed that the

“pre-irrigation” values of the indicators were the

same in the treatment and controls. However,

the irrigated areas tend to be more developed than

the unirrigated, having more education, better ac-

cess to markets, and a lower share of short-term

consultants. Hence, these areas may be expected

to have higher yields even in the absence of irri-

gation, so the single-difference estimate has an up-

ward bias.

Second, the estimates assume that irrigated areas

are entirely irrigated and unirrigated areas entirely

not irrigated. In fact, not all the irrigated area

will use canal waters. For this reason, the single-

difference estimate underestimates the impact of

irrigation. Third, the comparison area will in-

clude plots (probably the majority) using ground-

water irrigation. Bringing surface irrigation will

either replace one source with another or allow

conjunctive use of irrigation waters—the impact

of either of these being less than that of moving

from rain-fed to irrigated farming. Thus, the

single-difference estimate is an overestimate of

the likely actual impact of canal irrigation being

introduced. In summary, two of the biases are up-

ward (an overestimate) and one is downward

(an underestimate). (There is also the upward bias

from using the treatment of treated effect; see

appendix A).

By making some assumptions it is possible to

correct the biases. The second bias is most

straightforward, because the additional infor-

mation required is the share of the area that is ir-

rigated in the treatment districts. Data from the

baseline survey put this between 60 and 70 per-

cent; the lower figure is taken here, which gives

the larger increase in yields. Assuming that yields

in the unirrigated part of the treatment group are

the same as those for the comparison group,

then, for example, the 15 percent increase in

kharif paddy yields becomes 25 percent. Deal-

ing with the other two sources of bias requires

further assumptions: (1) that the social and eco-

nomic advantages give a pre-intervention yield

differential of 20 percent and (2) that the bene-

fit of going from groundwater irrigation to hav-

ing access to surface irrigation is 60 percent of the

benefit of going from rain-fed to surface irrigation,

and that this adjustment applies to 50 percent of

irrigated plots (the other 50 percent going from

rain-fed to canal irrigation).

These assumptions are applied in table C.8. The

unadjusted estimate is the average of the ICR es-

timate and the baseline survey estimate (calculated

as an average across all available years). The lower

estimate comes from correcting the sources of up-

ward bias and the upper estimate from correct-

ing that for downward bias. The point estimate


7 1

Table C.7: Plot-Level Single-Difference Estimates from 2006 Survey Data

Single differenceTubewell

Level Kg/ha Percent orUnirrigated Canal Tank Canal Tank Canal Tank borehole

Paddy (kharif) 3.7 3.0 5.6 –0.7 1.9 –19 51 4.6

Paddy (rabi) 1.2 4.9 3.0 3.7 1.8 308 150 2.5

Cotton 1.2 1.2 2.0 0.0 0.8 0 67 1.2

Chilies 0.5 0.7 0.8 0.2 0.3 40 60 1.0Source: IEG survey.Note: ha = hectares; kg = kilogram.

results from applying all three corrections at once,

these adjusted point estimates being slightly

higher than the unadjusted estimates. Consider-

ing the upper and lower estimates, the wide range

in these figures is not surprising. What is most

striking is that for the most important crop—

paddy—the largest estimate of the increased yield

is still less than one-third, and it may be less than

10 percent. These increases are considerably less

than those used at appraisal.

There are two further means to attempt to re-

move the biases in the single difference esti-

mate. The first is to ensure a valid match, that

is, ensure that pre-intervention conditions

were the same and trends similar other than

the presence of the intervention. The second

is to use double differencing to remove the pre-

intervention difference.

Controlling for pre-intervention differences can be

done in two ways. The first is to consider control

at the community level, which was done using the

method of propensity score matching described

in appendix A. As explained in that appendix, the

18 villages were placed in three groups: no canal

irrigation, new canal irrigation between the two

surveys, and existing canal irrigation at the time

of the first round. The tables show the single dif-

ference (changes between the two rounds) for

these groups. The no-irrigation category includes

tail-end villages intended to receive irrigation that

have not actually received it. This approach in

principle ensures that communities have com-

mon characteristics with regard to education,

caste composition, and market access, all of which

may affect agricultural productivity. However, as

seen in appendix A, making the match proved

somewhat difficult, given the fact that villages to

be irrigated are rather different from those al-

ready receiving irrigation. The single-difference

village-level estimates are given by the figures in

the 2006 column headed “Single difference” in

table C.9, which compares 2006 yields in villages

newly connected to the canal system with those

without such access in either year.

7 2


Table C.8: Single-Difference Yield Impact Estimates after Adjusting for Sources ofBias (percent increase in yield)

Adjusted

Unadjusted Kharif Rabi

Lower Point Upper Lower Point UpperKharif Rabi estimate estimate estimate estimate estimate estimate

Paddy 16 19 8 17 27 9 20 32

Maize 16 26 8 17 27 12 28 43

Cotton 38 n.a. 18 41 63 n.a. n.a. n.a.

Groundnut n.a. 50 n.a. n.a. n.a. 24 53 83Note: n.a. = not applicable.

Table C.9: Community-Level Yields by Project Irrigation Status (kg 000/ha)

2005 2006 Single Change (matched plots)

None New Existing None New Existing difference None New Existing

Paddy (kharif) 2.9 2.5 1.9 3.2 4.9 3.0 1.7 1.2 0.6 1.5

Paddy (rabi) 1.7 1.9 0.7 2.5 4.6 3.7 2.1 –0.7 0.6 1.8

Cotton 0.8 0.8 0.7 1.2 1.8 1.3 0.6 0.1 0.7 0.4

Chilies 0.6 0.4 0.1 1.0 2.0 2.0 1.0 0.5 2.0 0.1Source: IEG survey.Note: ha = hectares; kg = kilogram.

The community-level data can also be used for be-

fore versus after for newly irrigated plots or com-

munities. The average for all plots can be used,

or the sample can be restricted to those growing

that crop in both 2005 and 2006, thus controlling

for ecological conditions (such as soil type). This

is the figure shown in the last three columns of

table C.9, headed “Change (matched plots).”

This “before versus after” approach gives ab-

solute yield increases of 0.7 tons/ha (both sea-

sons), 0.6 tons/ha for cotton, and 2 tons/ha for

chilies; in percentage terms these figures are 24

and 32 percent for paddy in the kharif and rabi

seasons, respectively. They are much higher for

cotton (88 percent) and chilies (500 percent),

though the latter figure is based on too few ob-

servations to be a useful estimate. Before versus

after estimates are overestimates if there is a

general rising trend in yields in nontreatment

areas. Table C.9 shows this clearly was the case,

so these single differences are upper estimates

(and broadly consistent with the upper estimates

shown in table C.8, calculated by adjusting for the

bias in the treatment-comparison group single-

difference data discussed above).

The match can also be improved by controlling

for soil type. Sample sizes are too small to do

this for any crop other than paddy, and the com-

parison has to be for canal and tank irrigation

versus groundwater irrigation. The results, shown

in table C.10, give estimates of increased yields

ranging from 11 to 45 percent.

Measuring Impact Using DoubleDifferencesThe double-difference impact estimate is the dif-

ference in the change in the indicator of interest

between the treatment (project) and untreated

(comparison) areas, or, equivalently, the change

in the difference.

Using the first definition, the single difference is

the before versus after comparison of the indicator

in the project area. Before versus after does not

yield a reliable measure of impact because other

factors will have influenced the outcome during

the intervention. This bias is removed by the dou-

ble difference, which compares the change in

the project area with the change in the compar-

ison area, where the comparison area has been

subject to the same trends, with the important ex-

ception of the intervention being absent.

Using the second definition, the single difference

is the estimate reported in the previous section,

that is, the postintervention difference between

the values of the indicator in the treatment and


7 3

Table C.10: Single Difference Paddy Yields by Soil Type, 2006 (kg 000/ha)

Canal and Single difference

tank Groundwater None kg/ha Percent

All soils

Kharif 5.1 3.6 4.6 1.5 29Rabi 3.5 3.0 2.0 0.5 14

Red sandy soils

Kharif 6.5 3.6 n.a. 2.9 45Rabi 2.8 2.6 n.a. 0.2 7

Red soils

Kharif 5.5 3.0 n.a. 2.5 45Rabi 3.3 2.0 n.a. 1.3 39

Black chouduKharif 4.6 4.1 n.a. 0.5 11Rabi 4.5 3.4 n.a. 1.1 24

Note: ha = hectare; kg = kilogram; n.a. = not applicable.

comparison areas. This approach can be subject

to bias if the preintervention values of the indi-

cator were not the same. Hence, the double dif-

ference takes this into account by comparing

changes between project and comparison groups

rather than levels.

The potential problem with double-difference es-

timates is the presence of selection bias if bene-

ficiaries are selected conditional on variables that

are correlated with the indicator under exami-

nation. If these selection variables are observ-

able, then this bias can be removed by including

them as conditioning variables in the analysis, ei-

ther using a regression approach or matching

observations based on these characteristics. If se-

lection is on unobservables, then the bias cannot

be removed, although various approaches might

still be used to deduce impact by producing upper

or lower estimates if reasonable assumptions

might be made to calculate the magnitude of the

bias, as was done for the single-difference esti-

mates above.

Approach to double-difference estimates in this studyAs described in appendix A, matching for this

study was done at the community level. The

double-difference estimates are constructed us-

ing panel data from the IEG survey at the level of

the plot, the household, or the community. Table

C.11 summarizes the data. The discussion of yields

uses the plot and community-level panels.

The first (or single difference) of the double-

difference estimates can be made in two ways.

The first means of calculating single difference is

to use the 2006 data to compare treatment and

control areas, such as reported in the first section

of this appendix. At the plot level this approach

is the source of the figures in the column headed

“Single difference” in table C.7 and discussed

above. The community-level results were shown

in table C.9. The alternative single difference is the

before versus after, also discussed above.

As described above, the double difference is the

change in the differences, or difference in the

changes. The approach can be biased if the treat-

ment group differs systematically from the con-

trol with respect to variables that are also

correlated with the outcome. But the bias can be

removed by controlling for the selection vari-

ables, which this study does in the selection of

the communities to be included in the control

group. Fixed effects regressions are also used

for some outcomes.

But there is a further complication in making

double-difference estimates for agricultural vari-

7 4


Table C.11: Classifications According to Irrigation Status

Irrigation status Plot Household Community

None

New

Existing

Plot does not receivecanal or tank irrigationwater

Plot received canal ortank irrigation water forfirst time in kharif 2005

Plot received canal ortank irrigation waterprior to kharif 2005

No plots cultivated by the householdreceive canal or tank irrigation water

At least one plot cultivated by thehousehold received canal or tankirrigation water for first time inkharif 2005

At least one plot owned by thehousehold received canal or tankirrigation water prior to kharif 2005

Community either not connectedto SRPS system, or tail-endvillage officially connected butnot receiving water

Community received water forfirst time in kharif 2005

Community received water priorto kharif 2005

ables, which are subject to annual variations be-

cause of weather conditions. Harvests may be

broadly classified as poor, normal, or good, re-

flecting bad, average, or good years, respectively.

Hence, the year-to-year comparison may cover any

one of nine possible combinations (bad-bad, bad-

good, and so forth).

This creates two problems. First, the impact of ir-

rigation differs between bad and good years; it is

expected to be more in bad years, as large-scale

irrigation ensures water supply, which is less avail-

able to those depending on groundwater sources.

By contrast, in a good year water is plentiful, so

canal access matters less. (These observations

need to be further caveated by season: they apply

to the kharif season, in the survey areas rabi

crops, and summer crops are irrigation depend-

ent). So observations are needed across different

types of year to average the effect. The second

problem is that 2005 was not a bad year but a dis-

astrous one, in which no water was officially re-

leased in the SRSP command. Hence the “already

irrigated” sample did not receive water that year

(though a small number do claim to have done so).

Double-difference estimates

YieldYield estimates were calculated at the plot level,

with plots matched by both plot and season.

Hence, only those plots having the same crop, dur-

ing the same season, in consecutive surveys are

used in the analysis. The project group is identi-

fied in two ways: those plots using a new irriga-

tion source in 2006 (calculated separately for

canal, tank, tube, and borehole) and those plots

using a new irrigation source in 2006 that did not

have access to irrigation in 2005. The comparison

group for the first treatment group consists of

those plots with no change in irrigation source be-

tween 2005 and 2006. The comparison group for

the second treatment group comprises plots with

no irrigation in both 2005 and 2006.

These calculations are made for the four major

crop categories: paddy, cotton, chilies, and

turmeric. The results are shown in table C.12. In

some cases, the sample size of the panel is too

small, so the difference reported is that between

the unmatched group means. In other cases, the

sample size is too small for any estimates to be

made. Hence, although Control 2 is more ap-

propriate, Control 1 has to be used. Table C.13 re-

ports the double-difference estimates of cropping

intensity. The impact of canal irrigation is toward

the upper end of those reported above, though

the impact of tanks is lower and, as explained

elsewhere, tank irrigation picks up some of the

effect of canal water.


7 5

Table C.12: Double Difference Estimates of Impact on Yield (kg 000/ha)

Single difference Double difference

Control 1 Control 2 Control 1 Control 2

Paddy

New canal 1.3 1.4 0.7 1.0

New tank 1.4 1.7 0.7 1.4

New bore 1.0 1.7 0.3 1.8

New tube 1.4 1.6 0.7 1.3

None 0.7 0.3 n.a n.a

Cotton

New canal 0.7a . . 0.7 . .

New tank 0.6 1.0 0.6 0.7

New bore 0.4 0.7 0.4 0.4

New tube 0.5a 1.0 0.5 0.7

None 0.2 0.2 n.a n.a

Chilies

New canal 0.4a . . 0.2 . .

New tank 0.7a . . 0.4 . .

New bore 0.6a 0.7a 0.4 0.0

New tube 1.4a 1.7a 1.2 0.9

None 0.2 0.7a 0.0 0.0

Turmeric

New canal . . . . . . . .

New tank 1.6a . . 1.0 . .

New bore 1.2a . . 0.6 . .

New tube 1.2 . . 0.7 . .

None 0.6 . . 0 . .Source: IEG survey.Note: . . = not calculated owing to small sample size.a. Based on difference in simple (unmatched) group means (because of small panel sample size).

The impact estimates are calculated as follows:

(1) paddy—simple average of canal and tank

double difference using Control 2; (2) cotton—

simple average of tank double difference using

Control 2, and twice the Control 1 estimate for

canal, on the grounds that the Control 2 esti-

mates are approximately twice those of Control

1; and (3) turmeric and cotton—simple average

of canal and tank double difference using Control

1. The choices for making the estimates err on the

side of overestimating the benefits. But as noted

in the main text, estimated impact is on average

only half that expected at appraisal.

It should be mentioned that the yield impact es-

timates in I&CADD’s project completion report

(I&CADD 2004) are higher, averaging around

100 percent for paddy in Karimnagar and Waran-

gal districts; the season is not stated by the date

of data collection (March–May), which suggests

it may be rabi when the impact is higher. How-

ever, the data in that report are suspect because

they also discuss impact in the SRBC command,

7 6


Table C.14: Double-Difference Estimates of Household-Level Employment Effects

Domestic labor Employment

(days) Kharif (days) Rabi (days) Total

Single difference

New canal Control 1 –107 55 74 113

Control 2 . . . . 91 . .

New tank Control 1 –508 43 106 145

Control 2 –887 45 86 126

New tubewell Control 1 166 40 28 196

Control 2 311 38 34 82

New borehole Control 1 –827 20 30 75

Control 2 –640 10 42 89

None Control 1 –216 24 30 44

Control 2 140 5 19 31

Double difference

New canal Control 1 109 31 44 69

Control 2 . . . . . . . .

New tank Control 1 –292 19 76 101

Control 2 –1027 40 67 95

New tubewell Control 1 382 16 –2 152

Control 2 171 33 15 51

New borehole Control 1 –611 –4 0 31

Control 2 –780 5 23 58Note: . . = not calculated.

Table C.13: Double-Difference Impacton Cropping Intensity

Control 1 Control 2

New canal 41 . .

New tank 14 21

New bore 13 4

New tube 11 16

None 0 0Note: . . = not calculated.

despite the fact that no water had been released

there at the time the report was written. In the

case of paddy yields, the completion report cites

a 53 percent increase in “project villages” in

SRBC. There was no irrigation there, so this in-

crease can be combined with the 100 percent in-

crease in SRSP to give a double-difference yield

impact estimate of approximately 50 percent,

which is more consistent with the estimates used

in this study.

EmploymentDouble-difference employment effects were cal-

culated using the household panel by averaging

across the various irrigation categories (as used

for yields above). These results (table C.14) show

that adoption of canal or tank irrigation increases

employment throughout the year between 80

and 100 days. The fixed-effects regression (table

C.15) gives the same result.

The community-level panel can be used to con-

sider employment from the employee rather

than the employer perspective. The data show

that a new canal gives an increase in household

agricultural employment of about 80 days more

than that with no irrigation (table C.16). The

community-level analysis also shows wage ef-

fects. Wages, especially for women, rose across

the district in 2006, by most in areas with new

canal irrigation (table C.16).

Summary of SRSP EstimatesTable C.17 summarizes the various estimates of the

impact of irrigation on paddy yields. These results

are expressed in percentage terms, but it should

be remembered that it is the absolute increase that

matters for the rate of return analysis, because it

is the absolute increase in net farm income that

constitutes the core of the project benefits (es-

pecially for paddy, which has the largest crop

share). The “without project” value assumed in the

staff appraisal report (SAR) (3.2 tons/ha) is to-

ward the upper end of those observed in the dis-

trict—indeed at levels only recently attained in

the SRSP command itself (figure C.1). Table C.18

provides a summary of estimates for cropping

intensity.


7 7

Table C.15: Fixed-Effects Regression of EmploymentEffects

Kharif Rabi Total

Canal 33.8 27.1 60.9

–1.0 –1.1 –1.3

Tank –22.6 –41.3 –64.0

–1.1 (2.76)*** (2.28)**

Tubewell 1.6 22.5 24.2

–0.1 –1.4 –0.8

Borehole 24.3 48.0 72.4

–1.3 (3.56)*** (2.86)***

Canal* area –2.0 4.8 2.8

–0.6 (2.03)** –0.6

Tank* area 14.3 24.1 38.4

(4.73)*** (10.77)*** (9.17)***

Borehole* area –1.2 –11.8 –13.0

–0.4 (5.41)*** (3.17)***

Tubewell* area 1.6 –9.4 –7.9

–0.5 (4.26)*** (1.90)*

Constant 78.8 18.1 96.8

(6.68)*** (2.08)** (5.94)***

Observations 916 916 916

R-squared 0.32 0.09 0.43

Marginal effect at mean area

Canal 20 61 81

Tank 37 60 97

Tubewell 11 –37 –26

Borehole 9 –21 –12Note: Absolute value of t statistics in parentheses.

*Significant at 10 percent.**Significant at 5 percent.

***Significant at 1 percent.

Table C.16: Change in Employment(days/household/year) and Wages (Rs/day)

Change in Change in wages

employment Female Male

None 33 2.1* 0.2

New 114* 3.3* –1.9

Existing 42* 1.8 2.9*Source: IEG survey.*Significant at 5 percent.

Srisalam Right Bank CanalBecause the SRBC command has not yet received

irrigation, there are no ex post data available for

impact analysis. The approach adopted here is

that appraisal estimates are overly optimistic, so

that lower percentage increases are assumed.

Table C.19 shows the appraisal and ICR estimates,

together with those used in this study. The ini-

tial yields used are usually the lower of the esti-

mates from the two sources. In the case of paddy,

the lower of the two estimates (from the SAR) is

still the highest level achieved in Kurnool district

in the last 15 years; the average from 1989–90

to 2003–04 has been 2.6 tons/ha, peaking at 3.2

tons/ha in 2000–01 (data from the Directory of

Economics and Satistics, Andhra Pradesh).

7 8


Table C.17: Summary of Impact on Rice Yields (percentage increase)

Kharif RabiSingle Double Single Double

difference difference difference difference

ICR 15 n.a. 19 n.a.

IEG Mandal-level 63 n.a. 37 n.a.

“Natural experiment” 63 14 34 66

Community level 53 84 84 135

Community level (matched plots) 29 –24 14 n.a.

Plot-level 32 24 46 33

AFC (SRSP baseline) 16 n.a. 20 n.a.

Adjusted single difference 17 n.a. 20 n.a.

Memo: assumed in SAR 69 n.a. 69 n.a.Notes: ICE = Implementation Completion Report; n.a. = not applicable; SAR = Staff Appraisal Report.

Figure C.1: Yields in the SRSP Command Area

500

01976–77 1980–81 1984–85 1988–89 1992–93 1996–97 2000–01

Kharif

Rabi

2004–05

1,000

1,500

2,000

2,500

3,000

3,500

4,000

Yiel

d (k

g/ha

)

Years

Source: I&CADD and Directorate of Economics and Statistics.

A P P E N D I X C : S I N G L E A N D D O U B L E - D I F F E R E N C E I M PA C T E S T I M AT E S

7 9

Table C.19: SRBC Yield Impact Estimates

SAR ICR This study Percentage change

ThisWOP WP WOP WO WOP WP SAR ICR study

KharifPaddy 3.2 5.4 4.5 5.1 3.2 3.8 69 13 20

Cotton 1.1 2.5 1.0 1.5 1 1.5 127 50 50

Sorghum 1.2 2.5 1.0 2.0 1 1.6 108 100 60

Sunflower 0.6 1 0.6 1.2 0.5 0.8 67 100 67

Groundnut 1 2.2 1.0 2.0 1 1.6 120 100 60

Chilies 0.9 2.5 n.a 1.5 0.8 1.6 178 n.a. 60

RabiPaddy n.a. n.a. 4.0 5.5 3.6 5.0 n.a. 38 40

Groundnut 1.0 2.5 1.5 2.5 1.0 1.7 150 67 67

Sorghum 1.2 3 1.5 n.a. 1.2 2.0 150 n.a 67

Sunflower 0.6 2 1.0 1.8 0.6 1.1 233 80 80

Chickpea 0.6 1.5 1.0 2.0 0.6 1.1 150 100 80Source: IEG data.Notes: n.a. = not applicable; SAR = Staff Appraisal Report; WOP = without project; WP = with project.

Table C.18: Estimates of Impact on Cropping Intensity

Single difference Double difference

ICR 15 n.a.

IEG Mandal level 23 n.a.

“Natural experiment” 17 10

Plot level 14 21Note: n.a. = not applicable.

8 1

APPENDIX D: DATA TABLES

Table D.1: Irrigation Spending and Potential Created in Andhra Pradesh

Amount spent Irrigation potential created(million Rs.) (million ha)

Major and Major andmedium Minor Total medium Minor Total

Preplan period 1951 — — — 1.33 1.37 2.70

First plan (1951–56) 375 35 410 0.08 0.03 0.10

Second plan (1956–61) 574 44 618 0.18 0.02 0.20

Third Plan (1961–66) 915 186 1,101 0.37 0.05 0.42

Three annual plans (1966–69) 609 108 717 0.08 0.04 0.12

Fourth plan (1969–74) 1,187 182 1,369 0.19 0.06 0.25

Fifth plan (1974–78) 2,691 388 3,079 0.21 0.09 0.31

Two annual plans(1978–80) 2,577 238 2,815 0.15 0.06 0.21

Sixth plan (1980–85) 7,296 411 7,707 0.31 0.08 0.39

Seventh plan (1985–90) 13,064 1,314 14,378 0.09 0.07 0.16

Annual plan (1990–91) 2,828 478 3,305 0.01 0.01 0.02

Annual plan (1991–92) 3,339 486 3,825 0.01 0.01 0.02

Eighth plan (1992–97) 22,653 1,869 24,522 0.04 0.04 0.08

Ninth plan (1997–2002) 50,272 7,757 58,029 0.58 0.03 116,058.61

Tenth plan (2002–07) 91,538 16,072 107,610 n.a. n.a. 0.94Note: n.a. = not applicable; — = not available.

8 2


Table D.2: SRSP: Actual Area Irrigated as a Percentage of Planned Area Irrigated

1995–96 1996–97 1997–98 1998–99 1999–2000

Above Lower Manair Dam

Kharif

Head 129 94 140 87 85

Middle 55 74 78 49 85

End 45 40 68 88 88

Rabi

Head 55 76 . . 51 42

Middle 52 57 . . 66 n.a.

End 50 52 . . 67 99

Below Lower Manair Dam

Kharif

Head 25 44 59 66 73

Middle 42 38 55 113 110

End 50 51 53 97 89

Rabi

Head 26 41 . . 55 94

Middle 45 44 . . 98 112

End 40 42 . . 76 n.a

Simple average 51 54 75 76 88Source: AFC SRSP baseline survey, pp. 127–29.Note: . . = no water released this season; n.a. = not applicable.

Table D.3: Area under Different IrrigationSources as Percent of Total Area

2005 2006

Canal 4.6 5.2

Tank 7.8 20.5

Tube 44.8 50.5

Bore 27.0 26.5

None 23.6 11.3Source: IEG survey.Note: Sums to more than 100, as some plots have multiple sources.

A P P E N D I X D : D ATA TA B L E S

8 3

Table D.4: WUA Membership by Socioeconomic Characteristics (% households with WUA member), 2006

% households with member

CultivatingCategory All households households only

Landowning Landless 0.0 0.0

Marginal 18.4 18.4

Small 12.2 12.2

Large 28.2 28.2

Education None 8.4 13.7

Primary 20.7 30.1

Higher than primary 15.6 24.7

Caste Scheduled caste 13.3 28.5


Backward caste 13.2 21.8

Other caste 20.5 28.0

Wealth 1 0.8 2.0

2 7.6 11.4

3 9.3 13.7

4 28.0 37.0Source: IEG survey.

Table D.5: Gini Coefficients for Land Ownershipby Irrigation Source

Canal Tubewelland tank and borehole All land

2005 0.57 0.40 0.35

2006 0.45 0.39 0.34Source: IEG survey.

Staff appraisal report

Handbook of statistics Without WithCrop (2003–04) project project IEG survey

Paddy 2.2 3.2 5.4 3.8

Cotton 2.4 1.1 2.5 1.9

Chilies n.a. 0.9 2.5 1.6

Turmeric n.a. 3.2 4.0 2.0Sources: Handbook of Statistics Warangal District 2003–04; AP III ICR, appendix 10.Note: ha = hectare; kg = kilogram; n.a. = not applicable.

Canal Tank Tubewell Borehole All land

2005

1 4 10 7 7 8

2 5 10 23 14 19

3 23 9 28 24 26

4 69 71 43 54 47

2006

1 13 6 4 7 8

2 14 16 16 22 22

3 43 12 17 27 24

4 31 65 63 44 47

Total 100 100 100 100 100Source: IEG survey.

8 4


Table D.6: Share of Irrigated Land Owned by Wealth Quartile

Table D.7: Involvement in WUA Activities by Wealth Quartile (percent), 2006

Regularly attend Take part in Take part in InfluenceQuartile Member meetings O&M decision making water management

1 0.8 0.2 0.0 0.0 0.0

2 7.6 1.1 3.6 1.7 3.1

3 9.3 1.8 3.2 1.3 0.6

4 28.0 11.0 11.1 4.7 4.4Source: IEG survey.

Table D.8: Actual and Expected Crop Yields (kg 000/ha)


8 5

Table D.9: Crops Shares in Crop Income (percent)

% exclusiveRice Cotton Maize Chilies Turmeric Groundnut rice producers

2005

Irrigation source

Canal 61.7 24.7 0.5 6.8 2.1 2.7 42.1

Tank 61.2 16.8 6.1 2.2 2.9 0.8 48.1

Bore 37.9 36.5 2.9 8.5 5.3 2.7 17.4

Tube 42.1 36.2 0.9 6.3 4.2 3.1 20.0

None 32.1 32.6 1.7 6.5 3.5 1.2 22.0

Number of sources

0 32.1 32.6 1.7 4.5 0.7 1.2 22.0

1 41.2 35.0 2.2 7.3 4.3 2.7 22.3

2 49.2 27.4 4.9 6.0 6.0 2.4 22.9

Total 39.4 33.8 2.3 6.5 3.5 2.2 22.4

2006

Irrigation source

Canal 59.0 15.4 4.2 10.3 7.5 2.8 32.3

Tank 59.7 25.4 2.0 3.5 4.3 1.9 38.8

Borehole 44.5 30.2 2.0 8.0 6.7 2.2 17.4

Tubewell 42.4 28.9 2.8 8.4 8.0 5.5 18.5

None 43.6 34.2 4.8 1.0 1.0 2.2 20.0

Number of sources

0 43.6 34.2 4.8 1.0 1.0 2.2 20.0

1 48.6 27.7 2.4 7.4 5.0 3.0 23.7

2 47.8 27.1 2.3 7.2 11.1 2.7 21.5

Total 48.1 28.3 2.5 6.8 5.5 2.9Source: IEG survey.Note: Income is sales + imputed value of own consumption.

Cultivated area Gross Cropping(net cropped area) cropped area intensity

2005 1,610 2,414 150

2006 1,760 2,738 156

Percent increase 9.4 13.4 3.7Source: IEG survey.

8 6


Table D.10: Sources of Information on Agriculture-Related Matters

Received Source from which received information

information (as a percent of those who received information)

(percentage of Friendscultivating and Extension Community

households) family worker Training worker

2005

Use of fertilizer 7.7* 7.8 2.2 13.2* 74.7

Seeds 7.0* 8.5* 0.0 5.2 84.0

Pest management 0.9* 39.7 0.0 22.5 18.9

Soil conservation 1.0* 84.0* 0.0 16.0 0.0

Agricultural marketing 0.2 100.0 0.0 0.0 0.0

Irrigation methods 0.3 0.0 0.0 0.0 0.0

O&M of canals 0.0 n.a. n.a. n.a. n.a.

Water management 0.2 100.0 0.0 0.0 0.0

2006

Use of fertilizer 6.7* 50.4* 0.0 16.5 26.3

Seeds 3.4* 39.3* 0.0 32.0* 18.2

Pest management 1.5* 0.0 0.0 75.1* 0.0

Soil conservation 0.3 0.0 0.0 0.0 0.0

Agricultural marketing 4.1* 64.7* 0.0 6.2 18.7

Irrigation methods 0.4 0.0 0.0 100.0 0.0

O&M of canals 0.0 n.a. n.a. n.a. n.a.

Water management 0.0 n.a. n.a. n.a. n.a.Source: IEG survey.Note: n.a. = cannot be calculated as 0 positive responses to screening question; O&M = operations and maintenance.*Significant at 5 percent.

Table D.11: Gross and Net Cropped Area (hectares), 2005 and 2006

Canal Tank Bore Tube

Free intake 74 80 18 8

Manual pump 0 1 3 4

Electric pump 26 19 78 87

Other 0 0 1 1Source: IEG survey.


8 7

Table D.12: Cropping Intensity by IrrigationSource and Number of Sources

2005 2006

Irrigation source

Canal 202 199

Tank 148 151

Tube 159 168

Bore 143 169

None 139 90

Total 150 156

Number of sources

0 139 90

1 150 161

2 180 182Source: IEG survey.

Table D.14: Means of Water Transfer to Field by Irrigation Source, 2006

Table D.13: Average Crop Prices, 2005 and 2006 (Rs per quintal)

2005 2006

Mean Median Mean Median

Rice 718 672 605 581

Maize 929 643 511 500

Groundnut 1,063 977 1,420 1,447

Cotton 1,692 1,714 1,797 1,798

Chilies 1,430 1,441 2,272 2,129

Turmeric 1,011 606 1,605 1,550

Pulses 1,448 1,136 1,632 1,683

Jowar 2,000 2,000 900 900Source: IEG survey.

8 8


Table D.15: Agricultural Employment (average days per household)

Outside employment Own farm

All Cultivating Noncultivating All CultivatingYear households households households households households

2005 155 57 306 221 336

2006 223 157 313 190 317Source: IEG survey.

Table D.16: Average Employment of OutsideLabor by Season (days)

Season 2005 2006

Kharif 98 145

Rabi 2 46

Total 99 191Source: IEG survey.

Table D.17: Agricultural Employment by Gender (days)

2005 2006 Change

Kharif

Female 63 101 38

Male 37 34 –2

Rabi

Female 1 91 90

Male 1 54 54

Total

Female 64 199 135

Male 38 114 76

Percent female in total 63 64 64Source: IEG survey.Notes: Total is taken as total of subsample means. Differences caused by rounding.


8 9

Table D.18: Agricultural Employment by Wealth and Landholding (average days perhousehold), 2005

All Cultivating All CultivatingQuartile households households Landholding households households

1 278 88 Landless 306 n.a.

2 123 50 Small 79 79

3 159 70 Marginal 22 22

4 62 41 Large 21 21Source: IEG survey.Note: n.a. = not applicable.

Table D.19: Employment of Domestic Labor by Wealth and Landholding (averagedays per household), 2005


1 134 290 Landless n.a. n.a.

2 227 308 Small 273 273

3 239 348 Marginal 400 400


Table D.20: Agricultural Employment by Wealth and Landholding (average days perhousehold), 2006


1 294 209 Landless 316 n.a.

2 234 182 Marginal 223 210

3 239 168 Small 135 128


9 0


Table D.21: Employment of Domestic Labor by Wealth and Landholding (average days per household), 2006

All Cultivators All CultivatorsQuartile households only Landholding households only

1 82 265 Landless 0 n.a.

2 171 269 Marginal 244 265

3 223 338 Small 355 361


Table D.22: Agricultural Employment by Wealth and Land Ownership, 2005

Workers as percentage of Days worked as percent of those age 16–65 available days

Outside Outsideemployment Own farm employment Own farm

Wealth quartile

1 50 29 21 9

2 23 55 10 15

3 25 47 10 15

4 11 52 4 18

Landholding

None 57 n.a. 23 n.a.

Marginal 10 67 5 19

Small 3 80 1 23

Large 7 72 2 27

All 27 45 11 14Source: IEG survey.Note: n.a. = not applicable.


9 1

Table D.23: Agricultural Employment by Wealth and Land Ownership, 2006

Days worked as percent of Days worked as percent of available days available days

Outside Outsideemployment Own farm employment Own farm

Wealth quartile

1 73 23 24 7

2 59 41 18 12

3 53 49 17 15

4 29 46 9 19

Landholding

None 68 n.a. 25 n.a.

Marginal 55 59 15 18

Small 38 70 9 24

Large 20 65 6 27

All 53 40 17 14Source: IEG survey.Note: n.a. = not applicable.

Table D.24: Agricultural Day Wage (Rs/day)

2005 2006

Female Male Female Male

Mean 24.95 48.09 27.50 70.72

Median 25 50 30 50Source: IEG survey.Note: n.a. = not applicable.

9 2


Table D.25: Distribution of Wages by Year by Sex (Rs/day)

Female Male

2005 2006 2005 2006

<=20a 44 13 1 0

25 28 45 0 1

30 26 41 7 16

35 0 0 1 0

40 1 0 3 9

45 0 0 3 0

50 0 1 74 63

55 0 0 2 0

60 0 0 10 4

65 0 0 0 0

70 0 0 0 7Source: IEG survey.a. There is just one female observation below 20 in both years.

Table D.26: Socioeconomic Characteristics by Reach on Distributary Canal

Adult Scheduled Scheduled Pucca Distance toReach literacy caste (%) tribe (%) road (%) Warangal (km)

Head 48.1 15.7 20.1 67 39.1

Middle 45.6 14.1 24.3 74 53.0

Tail 43.6 15.7 18.8 65 76.4Source: Calculated from Andhra Pradesh census.

9 3

The farm model, based on that in the SAR, shows

farm income without project (WOP) and with ir-

rigation (with project, WP) for a typical hectare.

Farm models are estimated separately for SRBC and

SRSP. The basis for the yield increases and crop

shares is explained in appendix C. Farm inputs are

assumed to increase with irrigation, though not by

as much as assumed in the SAR. These increases

in inputs as a result of irrigation are confirmed by

the multivariate analyses (tables B.6–B.10). The

smaller increase than that expected at appraisal can

be linked to the reduced delivery of extension ser-

vices. Lower inputs are one reason that actual

yields are lower than those anticipated at appraisal.

The farm budget is for a typical hectare, with the

net income calculated as a weighted average of net

income from each crop (the weights being the

share in gross cropped area), the total multiplied

by cropping intensity. The resulting net farm in-

comes with and without the project are shown in

table E.1. In the case of SRSP, the IEG estimate lies

between those from the SAR and the ICR. But the

estimated increase in net farm income in the SRBC

command is less than half that calculated by the

other two reports. As explained at various points

in this report, there is no irrigation yet in the SRBC

command. Hence, the ICR relied on “informed

judgments” of the sort used in the SAR, presum-

ably explaining the similar estimate. But the figure

seems somewhat optimistic, being based on yield

increases larger than those observed elsewhere

and a diversification into high-value crops, which

has not occurred in other places (see appendix B).1

The benefits stream in the rate of return analysis

is net farm income (valued at economic rather

than financial prices, though only the latter are

shown here) multiplied by the irrigated area.

Given problems in reaching tail-enders and evi-

dence of overestimation of irrigated area by

I&CCAD, it is assumed that the irrigated area

reaches 80 percent of that stated in the SAR and

ICR. This figure may prove optimistic for SRBC,

where the Bank has questioned the adequacy of

the water supply. The results (table E.2) show

the low return to these investments. Estimates are

shown, including all investment costs for both AP

II and AP III, which is the relevant figure for this

report. Economic rates of return (ERRs) calculated

using only the AP III investment costs are also

shown. Further notes on the calculations will be

found in the tables beginning on page 97.

Poverty and Distributional Impact

Benefits per hectare of irrigationThe income benefits are the increment in farm in-

come and wage income to agricultural employees

(payment to domestic labor is already included in

APPENDIX E: THE FARM MODEL, RATE OF RETURN ANALYSIS,

AND DISTRIBUTIONAL IMPACTS

Table E.1: Net Farm Income (financial prices) With and Without Project (Rs/ha)

SAR ICR IEG estimates Percentage increase

WOP WP WOP WP WOP WP SAR ICR IEG

SRBC 7,469 3,4356 4,894 24,089 5,970 15,526 360 392 160

SRSP 9,796 22,792 8,842 14,965 7,524 16,070 133 69 114Note: ha = hectare; ICR = Implementation Completion Report; Rs = rupees; SAR = Staff Appraisal Report; WOP = without project; WP = with project.

net farm income; see table E.3). The increase in

employment income can be further broken down

into higher wages (benefiting all employees) and

increased employment. The additional employ-

ment is given by the labor demand in the farm

model; canal irrigation draws on additional ex-

ternal labor rather than household labor (a result

shown in both the double-difference and regres-

sion estimates). The distribution of this gain across

quartiles is based on the respective share of each

quartile in the incremental employment from

2005 to 2006 (table E.4). The benefit from higher

wages accrues to all those in employment and is

thus distributed across the quartiles according

to their share in 2006 employment. The whole of

the wage increase is being attributed to irriga-

tion, which is an overestimation, but these ben-

efits are relatively small, so this does not distort

the overall picture.

The allocation of net farm income is less straight-

forward, as average and marginal shares of irri-

gated land show a less clear pattern, although it

is clear that the bottom quartile has little and the

top quartile has the majority (table D.6). The cur-

rent distribution of farm income for the model is

the observed 2006 distribution, and the incre-

mental income share based on each quartile’s

share of canal and tank irrigated land in 2006.

The distribution of benefits from a hectare of ir-

rigation as a result of these calculations is shown

in figure E.1.

Community-level poverty modelTo calculate how benefits are distributed across

households at the community level, a model was

constructed using the 2006 survey data as the

baseline. The following assumptions were ap-

plied to the data:

Net farm income increases by Rs 8,800 (a weighted

average in the incremental net incomes for SRBC

and SRSP from the farm models) for each hectare

of land the household owns that will become ir-

rigated. The proportion of land irrigated is a func-

tion of income (one-fifth for the bottom quartile,

one-quarter for the second, one-third for the

third, and one-half for the top quartile).

The income of all households from agricultural

wages increased by 2.5 percent.

The gain from additional employment is calcu-

lated as a percentage increase in agricultural

SRBC SRSP

WOP WP WOP WP

Rs per hectare

Revenue 10,043 23,748 14,558 26,957

Costs 4,072 8,222 7,034 10,886

Net farm income 5,970 15,526 7,524 16,070

Wages 1,646 3,622 3,316 5,397

Days hired labor 88 188 177 280

Percentage increase

Revenue 136 85

Costs 102 55

Net farm income 160 114

Wages 120 63

Days hired labor 114 59Notes: ha = hectare; Rs = rupees; WOP = without project; WP = with project.

9 4


Table E.2: Estimates of the Economic Rate of Return

SAR IEG

All costs Excluding AP II ICR All costs Excluding AP II

SRBC 9 14 12 –2 0

SRSP 31 34 16 5 11

Whole project 19 24 15 2 6Note: SAR = Staff Appraisal Report.

Table E.3: Summary of Farm Model Results(average/ha)

labor income, where the increase is propor-

tionate to each quartile’s share in the marginal

increase in employment (table E.4), with an ad-

justment factor to capture the share attributable

to irrigation (the adjustment factor brings the

gains into line with those from the calculations

reported above).

The resulting income gains by quartile are shown

in figure E.2, which has a similar overall pattern

to figure E.1, the main difference being the smaller

benefit to the bottom quartile and larger benefit

to the top quartile. The model presentation breaks

down the wage and employment effects sepa-

rately. These two sets of results need not be the

same, as the former shows the distribution of in-

cremental benefits from irrigating 1 ha of land,

whereas the latter captures the effects of a com-

munity getting access to irrigation, allowing for dif-

ferential access to that irrigation.

Poverty impactThe simulation across the data set allows esti-

mation of poverty levels with and without irriga-

tion. Because the bottom quartile receives the

least benefits, the poverty impact is very sensitive

to the poverty line chosen (table E.5 and figure

E.2). The impact is also not that great because of

the finding that the bottom quartile does not

benefit from the marginal increase in employ-

ment. Relaxing that assumption and allocating

the employment benefit according to average

rather than marginal shares shows a slightly greater

poverty impact than for the lower poverty line

(scenario 2).2

The same data are used to calculate the Gini co-

efficients. These coefficients appear high because

of the large number of negative incomes (which

are set to zero for the calculation). Despite the

largest absolute benefit going to the top quartile,

the Gini is unaffected because relative income

growth is greatest among the second quartile—

or the lowest quartile, if they were to have a larger

share in the employment benefits (table E.6).

A P P E N D I X E : T H E FA R M M O D E L , R AT E O F R E T U R N A N A LY S I S , A N D D I S T R I B U T I O N A L I M PA C T S

9 5

Table E.4: Derivation of Share of Employment

Days of agriculturalemployment Share of

Quartile 2005 2006 Change change

1 278 294 16 0.06

2 123 234 111 0.40

3 159 239 80 0.29

4 62 132 70 0.25Source: IEG survey.

Figure E.1: Distribution of Benefits from 1 Hectareof Irrigation

500

01 2 3 4

Incremental wages


1,000

1,500

2,000

2,500

3,000

3,500

4,000

4,500

5,000

Rs p

er h

ecta

re

Quartile

Source: IEG survey.Note: Rs = rupees.

9 6


Table E.5: Poverty and Inequality Measures

Upper poverty line Lower poverty line

WP WPMeasure WOP WP scenario 2 WOP WP scenario 2

Poverty headcount 0.39 0.29 0.31 0.24 0.21 0.20

Poverty gap 0.22 0.21 0.21 0.18 0.18 0.18Source: IEG survey.

Table E.6: Quartile Income Shares and Income Growth

Income shares Percentage change

Quartile Base case Simulation 1 Simulation 2 Simulation 1 Simulation 2

1 4.4 4.1 4.3 8.6 17.5

2 13.5 14.0 13.7 15.5 12.6

3 21.9 21.9 22.0 10.7 11.0

4 60.2 60.0 60.0 10.3 9.9

Figure E.2: Distribution Benefits Across All Households by Wealth Quartile

01 2 3 4

1000

2000

3000

4000

5000

6000

7000

8000

Wealth quartile

Poverty Impact of Irrigation

Inco

me

gain

per

hou

seho

ld (R

s)

Incremental wages

Incremental employment


Pove

rty

mea

sure

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

0.40

0.45

Poverty headcount

Upper poverty line

Lower poverty line

Upper poverty line

Lower poverty line

Poverty gap

WOPWP

Note: Units of measure are poverty headcount and poverty gap. Rs = rupees; WOP = base case (no irrigation); WP = with project (irrigation).

AP

PE

ND

IX E

:T

HE

FA

RM

MO

DE

L, R

AT

E O

F R

ET

UR

N A

NA

LY

SIS

, AN

D D

IST

RIB

UT

ION

AL

IMP

AC

TS

97

SRBC: Per Hectare Financial Budget: Without Project—Kharif and Two-Season Crops

Sun- Seed Vege- Ground- Cotton Sugar- Existing NewPaddy Soybean Sorghum flower bandi tables nut Cotton seeds Chilies cane Turmeric Mulberry mango mango

Yield (tonnes/ha) 3.2 0.8 1.0 0.5 0.5 7.0 1.0 1.0 0.8 0.8 80.0 1.5 4.5 4.0 8.0Price 4,000 8,000 3,500 10,750 15,000 2,200 10,500 18,500 110,000 15,000 550 15,000 8,500 5,000 5,000Value 12,800 6,400 3,500 5,375 7,500 15,400 10,500 18,500 82,500 12,000 44,000 22,500 38,250 20,000 40,000ByproductsYield (t/ha) 1.6 0.0 2.0 0.0 0.0 0.0 1.0 0.0 0.5 0.0 0.0 0.0 0.0Price 300 120 300 20,400Value 480 240 300 10,200

Total gross value 13,280 6,400 3,740 5,375 7,500 15,400 10,800 18,500 92,700 12,000 44,000 22,500 38,250 20,000 40,000InputsSeeds (kg/ha) 40 75 8 10 15 0.8 90 13 3 4 8,000 3 1,250 0 10Price (/kg) 6.9 10 5 15 30 150 20 25 200 200 0.6 1,000 0.25 0 10Value 276 750 40 150 450 120 1,800 325 600 800 4,800 3,000 3,12.5 0 100Urea (kg/ha) 100 30 60 50 10 40 100 25 40 50 75 100 300 25 150Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 400 120 240 200 40 160 400 100 160 200 300 400 1,200 100 600DAP (kg/ha) 100 30 60 50 10 40 100 25 25 50 75 100 150 35 50Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 850 255 510 425 85 340 850 212.5 425 637.5 850 1,275 297.5 425MOP (kg/ha) 0 15 0 0 0 35 0 0 0 50 35 75 150 0 150Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 75 0 0 0 175 0 0 0 250 175 375 750 0 750Manure (t/ha) 2 1 0 0 0 4 2 0 2 3 6 10 4 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140 140 140 140 140Value 280 140 0 0 0 560 280 0 280 420 840 1,400 560 0 0Chemicals (kg/ha) 2.5 0 0 0 1 4 1 0 0 5 0 1 0 1 3Price (/kg) 252 252 252 252 252 252 252 252 252 252 252 252 252 216 216Value 630 0 0 0 252 1,008 252 0 0 1,260 0 252 216 648Labor (md/ha) 170 85 53 74 96 315 84 74 800 220 364 645 346 75 85Price/md 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5Labor costs (50%) 3,188 1,594 994 1,388 1,800 5,906 1,575 1,388 15,000 4,125 6,825 12,094 6,488 1,406 1,594Price/day 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 536 402 402 369 938 436 402 670 804 1,273 804 268 67 268

Total costs 6,461 3,470 2,186 2,565 2,996 9,207 5,593 2,427 16,710 8,284 14,851 19,175 10,853 2,087 4,385Net production value 6,819 2,930 1,554 2,811 4,505 6,193 5,208 16,073 75,990 3,716 29,150 3,325 27,397 17,913 35,615

98

AN

IMP

AC

T E

VA

LU

AT

ION

OF

IND

IA’S

SE

CO

ND

AN

D T

HIR

D A

ND

HR

A P

RA

DE

SH

IRR

IGA

TIO

N P

RO

JE

CT

S

SRBC: Per Hectare Financial Budget: Without Project—Rabi Season

Rabi Sunflower SorghumSunflower Sorghum seed seeds Vegetables Groundnut Coriander Chickpea Castor Tobacco

Yield (t/ha) 0.6 1.2 0.6 1.0 7.0 1.0 0.4 0.6 0.9 1.1Price 10,750 3,500 30,000 7,500 2,200 10,500 13,000 9,350 10,000 13,000Value 6,450 4,200 18,000 7,500 15,400 10,500 5,200 5,610 9,000 14,300ByproductsYield (t/ha) 0.0 2.4 0.5 0.0 0.0 1.0 0.0 0.0 0.0 0.0Price 120 10,750 300Value 288 5,375 300

Total gross value 6,450 4,488 23,375 7,500 15,400 10,800 5,200 5,610 9,000 14,300InputsSeeds (kg/ha) 10 8 10 12 0.8 90 12 50 5 0.5Price (/kg) 15 5 45 40 150 20 15 15 45 130Value 150 40 450 480 120 1,800 180 750 225 65Urea (kg/ha) 50 60 60 120 40 100 10 30 50 100Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 200 240 240 480 160 400 40 120 200 400DAP (kg/ha) 50 50 120 150 40 100 10 30 60 120Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 425 425 1,020 1,275 340 850 85 255 510 1,020MOP (kg/ha) 0 35 0 35 35 0 0 0 0 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 175 0 175 175 0 0 0 0 0Manure (t/ha) 0 0 3 0 4 2 0 1 3 1.5Price (/t) 140 140 140 140 140 140 140 140 140 140Value 0 0 420 0 560 280 0 140 420 210Chemicals (kg/ha) 0 0 2 0 4 1 25 2 3 1Price (/kg) 252 252 252 252 252 252 4 252 252 252Value 0 0 504 0 1,008 252 100 504 756 252Labor (md/ha) 74 53 90 90 315 84 36 34 92 166Price/md 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5Labor costs (50%) 1,388 994 1,688 1,688 5,906 1,575 675 638 1,725 3,113Total animal days (adj) 12 12 14 16 28 13 10 13 18 18Price/day 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 402 402 469 536 938 436 335 436 603 603

Total costs 2,565 2,276 4,791 4,634 9,207 5,593 1,415 2,842 4,439 5,663Net production value 3,886 2,212 18,585 2,867 6,193 5,208 3,785 2,768 4,561 8,638

AP

PE

ND

IX E

:T

HE

FA

RM

MO

DE

L, R

AT

E O

F R

ET

UR

N A

NA

LY

SIS

, AN

D D

IST

RIB

UT

ION

AL

IMP

AC

TS

99

SRBC: Per Hectare Financial Budget: With Project—Kharif and Two-Season Crops

Sun- Seed Vege- Ground- Cotton Sugar- Existing NewPaddy Soybean Sorghum flower bandi tables nut Cotton seeds Chilies cane Turmeric Mulberry mango mango

Yield (t/ha) 5.4 1.5 2.5 1.6 0.8 12.0 1.6 1.5 1.2 1.6 100.0 2.0 5.0 4.5 8.0Price 4,000 8,000 3,500 10,750 15,000 2,200 10,500 18,500 110,000 15,000 550 15,000 8,500 5.000 5.000Value 21,600 12,000 8,750 17,200 12,000 26,400 16,800 27,750 132,000 24,000 55,000 30,000 42,500 22.500 40.000ByproductsYield (t/ha) 2.7 0.0 5.0 0.0 0.0 0.0 1.0 0.0 0.8 0.0 0.0 0.0 0.0Price 300 120 300 20,400Value 810 600 300 16,320

Total gross value 22,410 12,000 9,350 17,200 12,000 26,400 17,100 27,750 148,320 24,000 55,000 30,000 42,500 22.500 40.000InputsSeeds (kg/ha) 40 75 12 10 15 0.8 90 3 3 4 8,000 3 1250 0 10Price (/kg) 6.9 10 5 15 30 150 20 25 200 200 0.6 1,000 0.25 0 10Value 276 750 60 150 450 120 1,800 75 600 800 4,800 3,000 312.5 0 100Urea (kg/ha) 100 75 100 70 30 200 60 90 100 200 275 175 250 60 125Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 400 300 400 280 120 800 240 360 400 800 1,100 700 1,000 240 500DAP (kg/ha) 150 60 50 80 20 120 150 150 250 100 100 150 150 35 50Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 1,275 510 425 680 170 1,020 1,275 1275 850 850 1,275 1,275 297.5 425MOP (kg/ha) 0 35 40 0 15 100 0 200 300 100 50 100 150 0 150Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 175 200 0 75 500 0 1,000 1,500 500 250 500 750 0 750Manure (t/ha) 3 2 2 2 2 6 3 3 4 6 8 15 5 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140 140 140 140 140Value 420 280 280 280 280 840 420 420 560 840 1,120 2,100 700 0 0Chemicals (kg/ha) 3 0.5 0 1.5 2 5 1.5 5 10 7.5 0 2 0 1 3Price (/kg) 252 252 252 252 252 252 252 252 252 252 252 252 252 216 216Value 756 126 0 378 504 1260 378 1,260 2,520 1,890 0 504 216 648Labor (md/ha) 200 100 90 80 110 380 100 225 1,240 300 420 680 330 50 60Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5Labor costs (50%) 3,850 1,925 1,733 1,540 2,118 7,315 1,925 4,331 23,870 5,775 8,085 13,090 6,353 963 1,155

Total animal days (adj) 25 16 14 13 11 30 16 20 28 36 40 29 8 2 8Price/day 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 536 469 436 369 1,005 536 670 938 1,206 1,340 972 268 67 268

Total costs 7,815 4,602 3,567 3,744 4,085 12,860 6,574 9,391 30,388 12,661 17,545 22,141 10,658 1,783 3846Net production value 14,596 7,398 5,784 13,457 7,915 13,540 10,526 18,359 117,932 11,339 37,455 7,860 31,842 20,717 36,154

10

0

AN

IMP

AC

T E

VA

LU

AT

ION

OF

IND

IA’S

SE

CO

ND

AN

D T

HIR

D A

ND

HR

A P

RA

DE

SH

IRR

IGA

TIO

N P

RO

JE

CT

S

SRBC: Per Hectare Financial Budget: With Project—Rabi Season

Rabi Sunflower SorghumSunflower Sorghum seed seeds Vegetables Groundnut Coriander Chickpea Castor Tobacco

Yield (t/ha) 1.1 2.0 1.0 1.0 12.0 1.7 0.5 0.8 0.9 1.1

Price 10,750 3,500 30,000 7,500 2,200 10,500 13,000 9,350 10,000 13,000

Value 11,825 7,000 30,000 7,500 26,400 17,850 6,500 7,480 9,000 14,300Byproducts

Yield (t/ha) 0.0 4.0 0.5 0.0 0.0 1.7 0.0 0.0 0.0 0.0

Price 120 10,750 300

Value 480 5,375 510

Total gross value 11,825 7,480 35,375 7,500 26,400 18,360 6,500 7,480 9,000 143,00InputsSeeds (kg/ha) 10 12 10 12 0.8 90 20 50 5 0.5

Price (/kg) 15 5 45 40 150 20 15 15 45 130

Value 150 60 450 480 120 1,800 300 750 225 65

Urea (kg/ha) 80 90 150 120 220 100 30 40 50 100

Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0

Value 320 360 600 480 880 400 120 160 200 400

DAP (kg/ha) 100 120 120 150 25 50 10 30 60 120

Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5

Value 850 1,020 1,020 1,275 212.5 425 85 255 510 1,020

MOP (kg/ha) 0 40 50 80 100 0 0 0 0 0

Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0

Value 0 200 250 400 500 0 0 0 0 0

Manure (t/ha) 2 2 3 3 6 3 0 2 3 1.5

Price (/t) 140 140 140 140 140 140 140 140 140 140

Value 280 280 420 420 840 420 0 280 420 210

Chemicals (kg/ha) 1.5 0 2 0 7 1 25 3 3 1

Price (/kg) 252 252 252 252 252 252 4 252 252 252

Value 378 0 504 0 1,764 252 100 756 756 252

Labor (md/ha) 90 70 90 80 380 105 40 40 92 166

Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5

Labor costs (50%) 1,733 1,348 1,733 1,540 7,315 2,021 770 770 1,771 3,196

Total animal days (adj) 13 14 14 16 34 18 10 13 18 18

Price/day 50 50 50 50 50 50 50 50 50 50

Animal costs (67%) 436 469 469 536 1139 603 335 436 603 603

Total costs 4,146 3,737 5,446 5,131 12,771 5,921 1,710 3,407 4,485 5,746

Net production value 7,679 3,744 29,930 2,369 13,630 12,439 4,790 4,074 4,515 8,555

AP

PE

ND

IX E

:T

HE

FA

RM

MO

DE

L, R

AT

E O

F R

ET

UR

N A

NA

LY

SIS

, AN

D D

IST

RIB

UT

ION

AL

IMP

AC

TS

10

1

SRBC: Per Hectare Financial Budget: Without Project

Kharif crops Two season Rabi crops

ExistingPaddy Maize Pulses Groundnut Cotton Chilies Turmeric mango Sunflower Paddy Maize Groundnut Pulses

Yield (t/ha) 3.0 2.0 0.6 0.6 0.8 0.8 2.5 4.0 0.6 3.0 2.2 1.0 0.6Price 4,000 3,800 10,500 10,500 18,500 15,000 15,000 5,000 10,750 4,000 3,800 10,500 10,500Value 12,000 7,600 6,300 6,300 14,800 12,000 37,500 20,000 6,450 12,000 8,360 10,500 6,300ByproductsYield (t/ha) 1.5 2.0 0.0 1.0 0.0 0.0 0.0 0.0 0.0 1.5 2.2 1.0 0.0Price 300 80 300 300 80 300Value 450 160 300 450 176 300

Total gross value 12,450 7,760 6,300 6,600 14,800 12,000 37,500 20,000 6,450 12,450 8,536 10,800 6,300InputsSeeds (kg/ha) 40 20 20 90 13 4 3 0 10 40 20 90 20Price (/kg) 6.9 10 10 20 25 200 1000 0 15 6.9 10 20 10Value 276 200 200 1800 325 800 3000 0 150 276 200 1,800 200Urea (kg/ha) 150 120 10 40 30 100 150 25 50 150 120 40 30Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 600 480 40 160 120 400 600 100 200 600 480 160 120DAP (kg/ha) 100 50 50 100 25 50 100 35 50 100 50 100 50Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 850 425 425 850 212.5 425 850 297.5 425 850 425 850 425MOP (kg/ha) 0 30 0 0 0 50 75 0 0 0 30 0 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 150 0 0 0 250 375 0 0 0 150 0 0Manure (t/ha) 2 3 0 2 0 3 10 0 0 2 3 2 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140 140 140Value 280 420 0 280 0 420 1400 0 0 280 420 280 0Chemicals (kg/ha) 2.5 2 0 1 0 5 1 1 0 2.5 2 1 0Price (/kg) 252 252 252 252 252 252 252 216 252 252 252 252 252Value 630 504 0 252 0 1260 252 216 0 630 504 252 0Labor (md/ha) 170 98 78 103 84 245 645 75 78 180 98 103 57Price/md 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5 37.5Labor costs (50%) 3,188 1,838 1,463 1,931 1,575 4,594 12,094 1,406 1,463 3,375 1,838 1,931 1,069

Total animal days (adj) 25 18 10 13 12 24 24 2 12 25 18 13 10Price/day 50 50 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 603 335 436 402 804 804 67 402 838 603 436 335

Total costs 6,661 4,620 2,463 5,709 2,635 8,953 19,375 2,087 2,640 6,849 4,620 5,709 2,149Net production value 5,789 3,141 3,838 891 12,166 3,047 18,125 17,913 3,811 5,602 3,917 5,091 4,151

10

2

AN

IMP

AC

T E

VA

LU

AT

ION

OF

IND

IA’S

SE

CO

ND

AN

D T

HIR

D A

ND

HR

A P

RA

DE

SH

IRR

IGA

TIO

N P

RO

JE

CT

S

SRBC: Per Hectare Financial Budget: With Project—Kharif and Two-Season Crops

ExistingPaddy Maize Sunflower Pulses Vegetables Groundnut Cotton Chilies Turmeric mango Soybean

Yield (t/ha) 4.2 2.2 1 0.9 12.0 1.2 1.6 1.6 3.5 4.5 0.5Price 4,000 3,800 10,750 10,500 2,200 10,500 18,500 15,000 15,000 5,000 8,000Value 16,800 8,360 10,750 9,450 26,400 12,600 29,600 24,000 52,500 22,500 4,000ByproductsYield (t/ha) 2.1 2.2 0.0 1.0 0.0 0.0 0.0 0.0Price 300 80 300Value 630 176 300

Total gross value 17,430 8536 10,750 9,450 26,400 12,900 29,600 24,000 52,500 22,500 4,000InputsSeeds (kg/ha) 40 20 10 20 0.8 90 3 4 3 0 75Price (/kg) 6.9 20 15 10 150 20 25 200 1,000 0 10Value 276 400 150 200 120 1,800 75 800 3,000 0 750Urea (kg/ha) 200 160 50 35 150 70 90 200 175 35 75Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 800 640 200 140 600 280 360 800 700 140 300DAP (kg/ha) 125 65 70 65 100 125 100 75 125 40 75Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 1,062.5 552.5 595 552.5 850 1,062.5 850 637.5 1,062.5 340 637.5MOP (kg/ha) 0 60 60 0 70 0 0 75 90 0 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 0 300 300 0 350 0 0 375 450 0 0Manure (t/ha) 2 3 3 0 0 2 0 3 10 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140 140Value 280 420 420 0 0 280 0 420 1,400 0 0Chemicals (kg/ha) 3.5 3 1 0.5 4 1 5 7 1.5 1.5 0.5Price (/kg) 252 252 252 252 252 252 252 252 252 216 252Value 882 756 252 126 1,008 252 1,260 1764 378 324 126Labor (md/ha) 205 110 88 65 300 110 180 245 710 75 80Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5Labor costs (50%) 3,946 2,118 1,694 1,251 5,775 2,118 3,465 4,716 13,668 1,444 1,540

Total animal days (adj) 25 18 16 10 25 15 18 30 26 2 10Price/day 50 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 838 603 536 335 838 503 603 1,005 871 67 335

Total costs 8,084 5,789 4,147 2,605 9,541 6,295 6,613 10,518 21,529 2,315 3,689Net production value 9,346 2,747 6,603 6,845 16,860 6,606 22,987 13,482 30,971 20,185 312

AP

PE

ND

IX E

:T

HE

FA

RM

MO

DE

L, R

AT

E O

F R

ET

UR

N A

NA

LY

SIS

, AN

D D

IST

RIB

UT

ION

AL

IMP

AC

TS

10

3

SRBC: Per Hectare Financial Budget: With Project—Rabi Season

ExistingChilies Turmeric mango Sunflower Paddy Maize Vegetables Groundnut Pulses Soybean

Yield (t/ha) 1.6 3.5 4.5 1.5 4.2 3.2 12.0 1.8 1.0 0.5Price 15,000 15,000 5,000 10,750 4,000 3,800 2,200 10,500 10,500 8,000Value 24,000 52,500 22,500 16,125 16,800 12,160 26,400 18,900 10,500 4,000ByproductsYield (t/ha) 0.0 0.0 0.0 0.0 2.1 3.2 1.8 0.0Price 300 80 300Value 630 256 540

Total gross value 24,000 52,500 22,500 16,125 17,430 12,416 26,400 19,440 10,500 4,000InputsSeeds (kg/ha) 4 3 0 10 40 20 0.8 90 20 75Price (/kg) 200 1,000 0 15 6.9 10 150 20 10 10Value 800 3,000 0 150 276 200 120 1800 200 750Urea (kg/ha) 200 175 35 50 200 60 150 70 35 75Price (/kg) 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0Value 800 700 140 200 800 240 600 280 140 300DAP (kg/ha) 75 125 40 50 125 65 100 125 65 75Price (/kg) 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5 8.5Value 637.5 1,062.5 340 425 1,062.5 552.5 850 1,062.5 552.5 637.5MOP (kg/ha) 75 90 0 0 0 60 70 0 20 0Price (/kg) 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0Value 375 450 0 0 0 300 350 0 100 0Manure (t/ha) 3 10 0 0 2 3 3 2 0 0Price (/t) 140 140 140 140 140 140 140 140 140 140Value 420 1,400 0 0 280 420 420 280 0 0Chemicals (kg/ha) 7 1.5 1.5 1 3.5 3 4 1 0.5 0.5Price (/kg) 252 252 216 252 252 252 252 252 252 252Value 1,764 378 324 252 882 756 1,008 252 126 126Labor (md/ha) 245 710 75 90 205 122 250 110 65 80Price/md 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5Labor costs (50%) 4,716 13,668 1,444 1,733 3,946 2,349 4,813 2,118 1,251 1,540

Total animal days (adj) 30 26 2 13 18 20 25 16 16 10Price/day 50 50 50 50 50 50 50 50 50 50Animal costs (67%) 1,005 871 67 436 603 670 838 536 536 335

Total costs 10,518 21,529 2,315 3,195 7,850 5,487 8,998 6,328 2,906 3,689Net production value 13,482 30,971 20,185 12,930 9,580 6,929 17,402 13,112 7,594 312

1 0 4


SRBC Economic Rate of Return (Rs)

Investment Addtitional Total Incremental NetYear costs O&M costs benefit benefit

1989 284 0 284 0 –284

1990 347 0 347 0 –347

1991 294 0 294 0 –294

1992 1,005 0 1,005 0 –1,005

1993 918 0 918 0 –918

1994 2,354 0 2,354 0 –2,354

1995 2,843 0 2,843 0 –2,843

1996 768 0 768 0 –768

1997 787 0 787 0 –787

1998 578 0 578 0 –578

1999 714 0 714 0 –714

2000 1,148 0 1,148 0 –1,148

2001 1,287 0 1,287 0 –1,287

2002 1,348 0 1,348 0 –1,348

2003 867 0 867 0 –867

2004 827 5 832 74 –758

2005 0 5 5 74 69

2006 0 16 16 263 247

2007 0 23 23 368 346

2008 0 33 33 547 514

2009 0 33 33 547 514

2010 0 33 33 547 514

2011 0 33 33 547 514

2012 0 33 33 547 514

2013 0 33 33 547 514

2014 0 33 33 547 514

2015 0 33 33 547 514

2016 0 33 33 547 514

2017 0 33 33 547 514

2018 0 33 33 547 514

2019 0 33 33 547 514

2020 0 33 33 547 514

2021 0 33 33 547 514

2022 0 33 33 547 514

2023 0 33 33 547 514

2024 0 33 33 547 514

2025 0 33 33 547 514

2026 0 33 33 547 514

2027 0 33 33 547 514

2028 0 33 33 547 514Note: ERR = –2 percent. Rs = rupees.

A P P E N D I X E : T H E FA R M M O D E L , R AT E O F R E T U R N A N A LY S I S , A N D D I S T R I B U T I O N A L I M PA C T S

1 0 5

SRSP Economic Rate of Return (Rs)


1989 3,581 0 3,581 0 –3,581

1990 431 0 431 0 –431

1991 380 0 380 0 –380

1992 633 0 633 0 –633

1993 694 0 694 0 –694

1994 890 0 890 0 –890

1995 1,245 0 1,245 0 –1,245

1996 728 16 744 216 –528

1997 1,643 16 1,660 216 –1,443

1998 1,209 49 1,257 649 –609

1999 1,456 65 1,522 865 –657

2000 1,863 81 1,944 1,081 –863

2001 2,623 90 2,713 1,189 –1,524

2002 1,156 98 1,254 1,297 43

2003 2,027 106 2,133 1,405 –728

2004 1,549 114 1,664 1,514 –150

2005 0 122 122 1,622 1,499

2006 0 130 130 1,730 1,599

2007 0 130 130 1,730 1,599

2008 0 130 130 1,730 1,599

2009 0 130 130 1,730 1,599

2010 0 130 130 1,730 1,599

2011 0 130 130 1,730 1,599

2012 0 130 130 1,730 1,599

2013 0 130 130 1,730 1,599

2014 0 130 130 1,730 1,599

2015 0 130 130 1,730 1,599

2016 0 130 130 1,730 1,599

2017 0 130 130 1,730 1,599

2018 0 130 130 1,730 1,599

2019 0 130 130 1,730 1,599

2020 0 130 130 1,730 1,599

2021 0 130 130 1,730 1,599

2022 0 130 130 1,730 1,599

2023 0 130 130 1,730 1,599

2024 0 130 130 1,730 1,599

2025 0 130 130 1,730 1,599

2026 0 130 130 1,730 1,599

2027 0 130 130 1,730 1,599

2028 0 130 130 1,730 1,599Note: ERR = 5 percent. Rs = rupees.

1 0 6


Combined SRBC and SRSP Economic Rate of Return (Rs)


1989 3,865 0 3,865 0 –3,865

1990 778 0 7,78 0 –778

1991 675 0 675 0 –675

1992 1,639 0 1,639 0 –1,639

1993 1,612 0 1,612 0 –1,612

1994 3,245 0 3,245 0 –3,245

1995 4,088 0 4,088 0 –4,088

1996 1,496 16 1,512 216 –1,296

1997 2,430 16 2,447 216 –2,230

1998 1,786 49 1,835 649 –1,187

1999 2,170 65 2,236 865 –1,371

2000 3,011 81 3,092 1,081 –2,011

2001 3,910 90 4,000 1,189 –2,811

2002 2,504 98 2,602 1,297 –1,304

2003 2,895 106 3,001 1,405 –1,595

2004 2,377 119 2,495 1,587 –908

2005 0 127 127 1,695 1,569

2006 0 146 146 1,993 1,847

2007 0 153 153 2,098 1,945

2008 0 164 164 2,277 2,113

2009 0 164 164 2,277 2,113

2010 0 164 164 2,277 2,113

2011 0 164 164 2,277 2,113

2012 0 164 164 2,277 2,113

2013 0 164 164 2,277 2,113

2014 0 164 164 2,277 2,113

2015 0 164 164 2,277 2,113

2016 0 164 164 2,277 2,113

2017 0 164 164 2,277 2,113

2018 0 164 164 2,277 2,113

2019 0 164 164 2,277 2,113

2020 0 164 164 2,277 2,113

2021 0 164 164 2,277 2,113

2022 0 164 164 2,277 2,113

2023 0 164 164 2,277 2,113

2024 0 164 164 2,277 2,113

2025 0 164 164 2,277 2,113

2026 0 164 164 2,277 2,113

2027 0 164 164 2,277 2,113

2028 0 164 164 2,277 2,113

Note: ERR = 2 percent. Costs are investment costs, excluding dam safety, O&M, and land costs. All prices converted to 2006 prices, using wholesale priceindex deflator where necessary. Incremental benefit is net farm income from farm model (valued at economic prices) times newly irrigated area. Standardconversion factor is 0.92. Rs = rupees.

1 0 7

APPENDIX F: VILLAGE PROFILES

10

8

AN

IMP

AC

T E

VA

LU

AT

ION

OF

IND

IA’S

SE

CO

ND

AN

D T

HIR

D A

ND

HR

A P

RA

DE

SH

IRR

IGA

TIO

N P

RO

JE

CT

S

Appendix F: Village Profiles

WUA member(% all Irrigation source (% of all plots)

households) 2005 2006

Village Water availability Canal condition Reach 2005 2006 Canal Tank Canal Tank

Uyyalawada No water Canal not completed because Tail 0 2 1.6 0.0 0.0 1.5of land dispute; heavy siltation

Vennaram Received water once only in 2004 Canal not completed Tail 12 4 4.1 0.0 4.0 0.0

Lingagiri No water Not yet constructed n.a. 2 0 0.0 9.6 0.0 23.6

Pragthisingaram No water Not yet constructed n.a. 2 14 0.0 10.3 5.0 62.5

Narsapur Water since rabi 2006 n.a. n.a. 2 20 0.0 11.4 13.6 44.1

Ingurthy Water since 2005; most goes to tanks Good Head 10 14 17.1 2.4 11.1 8.9

Komatipalli Water since 2005, goes to tanks preventing Main distributary Middle 18 10 2.9 1.4 7.0 8.5flow to two villages downstream clogged, tertiary good

Upparagudem Received water once only in 2004 Canal completed by Middle 0 12 0.0 18.9 2.6 15.4heavily silted

Kampalle Received water once only in 2004. Heavily silted Tail 0 8 0.0 0.0 0.0 13.0Water taken by farmers upstream

Nallabelli Water since rabi 2006 n.a. n.a. 10 28 2.4 14.3 4.0 44.0

Rajanpalli No water Not yet constructed n.a. 12 18 5.6 9.3 0.0 12.7

Malaya Water since 2003, flows to tanks Main distributary good, Middle 0 4 6.3 4.2 5.6 9.3tertiaries clogged

Mahadesapuram Received water once only in 2004 Heavily silted Head 0 0 0.0 2.0 0.0 1.9

Thopanpally Water since 2003, goes to tanks preventing Good Head 12 14 11.9 2.4 17.8 6.7flow to three villages downstream

Deekshakunta Water since 2005, flows to tank Good Middle 13 19 2.2 13.0 3.7 22.2

Peddakorpolu No water, diverted upstream Distributary unusable Tail 10 0 0.0 11.5 0.0 12.3to other villages

Chinnakondapak Water not released because farmers n.a. n.a. 0 20 0.0 26.8 0.0 30.4want to release to tanks to which ID opposed

Muchimpala Canal Good Middle 10 12 0.0 8.6 12.8 25.6Note: n.a. = not applicable.

1 0 9

I. Introduction1

Irrigation is definitely a powerful force for poverty

reduction. The expansion in the irrigation sector

and adoption of new techniques of agriculture

(Green Revolution) has resulted in an increase of

per capita grain production, which clearly shows

that the reason behind it is irrigation. Hence, ir-

rigation cannot be concluded as a paradox, and

it is the basic requirement for the sustenance of

mankind. Any developmental activity has got its

diversified impact, and a balanced view has to be

taken when an assessment is being taken. In-

vestments made toward irrigation infrastructure

are mainly one time, but the benefits (direct/

indirect) are long term. The emphasis on new

construction, rehabilitation of systems, and qual-

ity of work is altogether different from the

schemes planned especially to eradicate poverty.

It is agreed that the findings of the IEG study are

important and definitely provide altogether dif-

ferent perceptions, but to draw conclusions on the

entire irrigation development portfolio is in-

appropriate and not justifiable. The irrigation in-

frastructure creation as an effective mechanism for

fostering growth and economic upliftment in the

project areas is proved beyond doubt across the

globe. Moreover, the methodology of assessing the

impact, the sample size, and the sample chosen

by IEG has significant differences, and therefore,

the findings are misleading.

It is very much understandable that the two proj-

ects taken up are of different nature; that is, one

is basically a rehabilitation of an existing system

(SRSP) and the other one is set to complete a new

irrigation system for bringing new areas under

irrigation. Thus, special issues concerned with

these two projects and circumstances considered

by the Bank for assisting are overlooked by the

study, and the report is made exclusive of these

facts. It may not be out of context to mention that

IEG has prejudged the issues and extrapolated/

generalized the small sample survey results across

the irrigation investments.

However, the project-specific status with re-

marks and participatory irrigation management

efforts taken up by the state are highlighted

below so that the objective assessment of the

AP II and AP III projects is brought out in a more

focused manner.

II. Performance of SRBC and SRSPProjects

SRBC project Salient features of SRBC and the present progress

of the project:

• The SRBC scheme was formulated to irrigate

190,000 acres by utilizing 19 TMC of Krishna

water from the foreshore of Srisailam Dam to

benefit the chronic drought-prone areas in

Nandyal, Banaganapalli, and Koilakuntla taluks

of Kurnool district and Jammalamadugu taluk

of Kadapa district.

• The SRBC scheme was cleared by the plan-

ning commission in 1981.

• After completion of preliminaries, the works

from km 0–km 10 of SRBC were commenced

in 1984.

• Because of paucity of funds and to aid in the

speedy execution of the project, part of the

scheme was posed for World Bank assistance

in 1984.

APPENDIX G: COMMENTS FROM GOVERNMENT OF ANDHRA PRADESH

• 37 percent of the works were completed by the

time AP II was closed in 1994.

• The balance of the works of the project (except

Gorakallu balancing Reservoir, Owk second

stage, and main canal beyond 141.00 to 199.00

under SRBC scheme) was taken up with World

Bank assistance under a separate project—

AP III.

• Under AP II and AP III, all infrastructure to cre-

ate irrigation potential to an extent of 1,53,936

acres under 16 blocks was completed except

a micro-network.

Status of the Project: As of March 2004, Rs.

988.52 Crores had been spent, and the main canal

and distribution system were completed; as of

December 2006, 1,192.46 Crores had been spent

and water supplied through field channels to an

extent of 88,121 acres.

Works in Progress in SRBC: Sri Narasimharaya

Sagar Project—

1. Excavation of field channels:

(a) From Block I to X by Ayacutdar Committees/

water user associations (WUAs)

(b) From Block XI to XIV by Contracting agency

(c) Phase II works of Owk Reservoir Complex.

2. SRBC from km 141 to km 199, including dis-

tributaries XVII to XXI and micro-network.

Present Scenario: Expenditure particulars under

World Bank aid

Andhra Pradesh II irrigation project—333

Crores (approximately)

AP III irrigation project—845 Crores (approx-

imately)

(a) Micro-network under XV and XVI blocks

were completed.

(b) Micro-network under I to X blocks are

being executed under CADWM program.

(c) Micro-network under XI to XIV are being

executed by contracting agency.

(d) Irrigation potential created up to 2006

under I to XVI blocks is 153,936 acres and

ayacut brought under irrigation is 87,895

acres by constructing field channels.

(e) Since 2004, until water is supplied through

the SRBC system to an ayacut of 30,000 A/c

to 130,000 A/c as per the ayacut develop-

ment, as shown below (includes without

field channels also):

2004–30,000 A/c

2005–50,000 A/c against 160,000 A/c

2006–130,000 A/c }(f) Hence the benefits are now reaching the

beneficiaries. Moreover, the WUAs were re-

cently elected and have been functioning

since November 2006.

(g) The sample survey for SRBC will reveal a

different picture if taken up now as it is pro-

grammed to release water for entire aya-

cut of 160,000 A/c during kharif 2007.

(h) Furthermore, the CWC is conducting

monitoring every year and communicat-

ing its observations and the same are

being complied.

With regard to the sample selection of IEG irri-

gation survey, it has considered two command

areas in SRBC and SRSP. At the time of the first

round of the survey, very little water had been re-

leased, irrigating 7,800 ha of a planned 65,000 ha

in the SRBC command, and the simple reason

is that the micro-network was completed to an

extent of 20 percent. Completion of the network

was estimated for 2008, so IEG has not undertaken

the second round of a survey in 2006 in the

SRBC command. This is to say that 12 percent

of lands were able to be irrigated, and if no

system is developed, it could have been simply nil.

The deficiency of 8 percent cannot be directly

called a shortfall, as the farmers will take sufficient

time to develop their lands even after water is

made available to them, due to their financial

and other constraints. This is quite evident

when we observe the development of ayacut

in most successful schemes like Godavari Delta,

Krishna Delta, and the Nagarjuna Sagar Project.

Further, SRBC was a new scheme, whereas SRSP

was an extension of existing system under

AP II.

1 1 0


The government departments will be strength-

ened in due time for maintaining efficient man-

agement of the irrigation system, so as to redress

the weakness in the activities of WUAs.

IEG has come to a conclusion that the project’s fail-

ure can be attributed to problems both in design

and implementation. The rates that were adopted

in the estimates are purely based on realistic val-

ues and hence only the contractors are able to

execute the works with discounts. Further, con-

cluding that government officials at all levels ben-

efited financially from irrigation source, from the

study of Robert Wade (1982) in itself is not ac-

ceptable to the state and the same cannot be sub-

stantiated and the statement is not in good taste.

Sriramasagar projectSalient features of the project: The Srirama-

sagar project was constructed across the River

Godavari to create an irrigation potential of

968,640 acres in four districts: viz., Adilabad,

Nizamabad, Karimnagar, and Warangal. Before

taking up AP III, 630,000 acres of irrigation po-

tential was created. The area irrigated during

1995–96 was 268,301 acres, against the irrigation

potential creation of 630,000 acres.

The third Andhra Pradesh irrigation potential was

taken up to complete the irrigation development

and scheme rehabilitation works began under

second Andhra Pradesh irrigation potential and

thus realize the potential for increasing agricultural

production and rural income in backward regions

of Telangana, covering the districts of Adilabad,

Nizamabad, Karimnagar, and Warangal. Rehabili-

tation and modernization of Kakatiya Canal from

0 km to 234 km and distributory system to stabi-

lize the aycut of 625,000 acres.

District breakdown:

Nizamabad: 11,000 acresKarimnagar: 543,000 acresWarangal: 71,000 acres

Total: 625,000 acres

Objectives of AP III:

• Stabilization of 625,000 acres ayacut of Kakatiya

Canal up to 234 km

• Dam safety works (SRS Dam, Lower Manair

Dam) as recommended by Dam Safety Panel

• Environment Management Plan

• Agricultural support services

• R&R of project-affected persons

• O&M works.

Achievements:

The activities envisaged under AP III are imple-

mented. As a result the carrying capacity of

Kakatiya Canal increased from 8,250 cubic feet per

second (cusecs) to 9,700 cusecs. Thus the water

reached to tail-end areas of all the distributaries,

that is, from 0 to 234 km. The area irrigated also

increased from 268,301 acres to 796,887 acres

(2001–2002).

The following comments are offered on the Re-

port prepared by IEG on impact evaluation of AP

II and AP III projects in respect of Sriramasagar

Project:

• AP II could not be completed as planned mainly

because of delay in finalization of an R&R plan.

There was considerable delay in finalizing

model bid documents for civil works. The World

Bank indicated that it preferred to close AP II

instead of extending the project by 2–3 years.

• Appraisal for AP III, which commenced in Feb-

ruary 1994 (just prior to closure of AP II in June

1994), took three years.

• AP III became effective in July 1997. The clos-

ing date was extended by 18 months and the

project was closed in June 2004.

• Even though some problems in procurement

were encountered initially, AP III closed on a

successful note. IEG has made drastic com-

ments on the functioning of government de-

partments. It is felt desirable not to make an

issue of these sorts of comments.

• However, to put the issues in proper perspec-

tive the following points are deemed necessary.

The issue of changing to crops other than

paddy defies solution, as seen from past ex-

perience. It is better if this issue is not high-

lighted any longer. During the course of AP

III appraisal, it was hinted that supplies from

the canal should be restricted to ID crop

requirement. The farmers can supplement

A P P E N D I X G : C O M M E N T S F R O M G O V E R N M E N T O F A N D H R A P R A D E S H

1 1 1

balance requirement from ground water

or restrict their irrigated area.

Without having a first-hand knowl-

edge of problems encountered in im-

plementation of AP II and AP III, making

light of the work done achievements is

not fair.

Several supervision missions of Bank

study teams have visited the project dur-

ing implementation and appreciated the

work being done. The manner in which

the benefits of AP III are assessed and

quantified may differ from agency to

agency.

Verifiable benefits of civil works under AP III:

(a) Enhanced carrying capacity of Kakatiya Canal

from 8,250 cusecs to 9,700 cusecs. Records

maintained in the department will prove this

point.

(b) The distribution system that had deterio-

rated over a period of time was put back in

place. The area irrigated increased over a pe-

riod of time, with completion of rehabilitation

and modernization works. Reports of various

supervision missions contain the details.

(i) Before AP III, the maximum discharge in

to the main canal was to 6,000 cusecs ver-

sus 8,250 cusecs.

(ii) After taking up and completion of AP

III, the canal discharge was greater than

9,000 cusecs.

(c) The achievements listed in (a) and (b) above

were possible with the same quantum of

water. The inflows in to the reservoirs and re-

leases over several years can be verified.

(d) Productivity has considerably increased and

so has the income of farmers.

(e) Farmers’ organizations, which were put in place

in 1997 for a 5-year term, functioned fairly well.

However, it has to be put on record that the con-

clusions of the study do not adequately represent

ground reality. The conclusions are not based on

project-specific data.

III. General Remarks about Sample Size The survey consisted of two rounds in June 2005

and March 2006, with limited interrogation of 50

households in each village of the 16 total villages

covered and the data collected on income, farm

budgets, irrigation, and various aspects of social

and political life in the village cannot be formed

as a base, especially in the countries like India,

which is in developing stages. Thus, the conclu-

sions formulated based on the limited survey

with inadvertent public cannot form a base to

extrapolate the entire situation and thus the con-

clusions will not be realistic.

The study, which is a sample one, has revealed cer-

tain things. A sample can be called as a viable

sample if it satisfies the requirements of certain

principles. Further, conclusions made on SRBC

based on sample of SRSP are inappropriate.

IV. Participatory Irrigation ManagementTo improve irrigation performance, the govern-

ment of Andhra Pradesh took a progressive and

innovative step to empower the farmers to man-

age and operate the irrigation resources through

formation of WUAs in 1997. The exclusive Andhra

Pradesh Farmers Managed Irrigation Act of 1997

was enacted, which provides the legal support for

the functioning of these WUAs.

The objectives of farmers’ organizations are car-

rying O&M of the irrigation system, effective water

management, and increased agricultural pro-

duction. The act emphasized formation of water

users’ organizations at three levels, namely WUAs,

distributory committees (DCs) and project com-

mittees for major irrigation projects; two-tier

structure (WUA/PC) for medium irrigation proj-

ects; and single-tier structures (WUA) for minor

irrigation.

The following remarks are made on IEG’s obser-

vations regarding participatory irrigation man-

agement (PIM).

1. The expansion and durability of WUAs

has been influenced by external forces.

In 1997, WUAs were constituted for all the irri-

gation sectors and numbered around 10,000 for

involvement of the stakeholders in the man-

agement of irrigation systems. DCs were also

constituted.

1 1 2


After completion of the WUAs’ 5-year tenure, the

performance of these bodies is evaluated and

thought to bring changes in the set-up of the

WUAs for more accountability and more con-

centration on other statutory requirements, in-

cluding water management.

It is felt that the WUAs are to be made continu-

ous bodies to have a blend of old and new mem-

bers for better working of these organizations.

To bring the above changes, there was a gap of

about one year for conduct of elections to the

WUAs.

Elections to all the WUAs in the state were com-

pleted and elections for the DCs were also com-

pleted in 13 districts and will be completed by the

end of April 2007 in the remaining districts. WUAs

and DCs are also constituted in the SRBC project

during November 2006. Constitution of project

committees is also under active consideration of

the government.

2. WUA membership is not representative

of the population served.

IEG observed that half of the households in 18

villages of Warangal district have access to canal

or tank irrigation, but only 18 percent of the all

households are members. In this connection, the

Act provided membership only for the land hold-

ers; that is, land is to be in the name of the per-

son who has voting rights and who is a member

of the association. The tenant will get voting

rights and be a member of the association if the

land holder leases the land. Normally, the land

is registered in the name of the head of the fam-

ily; thus, the households are represented in a

smaller percentage.

Representation of the WUAs is mostly by the elite

groups; the study conducted by the IEG group is

only for a limited area and for all practical purposes

the study conducted by ORG (2005) can be taken

as reliable study, as the sample taken for the same

is 214 WUAs across the state and in all three sec-

tors of irrigation. Backward castes represented

around 60 percent of the managing committee of

the WUAs. The study also revealed that around 50

percent of the marginal farmers are represented

in the managing committee of the WUAs.

Thus, it is not correct that most of the managing

committees of WUAs are represented by elite

groups. In this connection, it is informed that all

the land holders are the voters of the WUA and

these are modeled as self-help groups; there are

no reservations based on social status. However,

in the future, the representation of small and

marginal farmers and also other social groups

will also increase over and above the present

level, as is happening in the other institutions.

Regarding benefits to the poor, the AP II and AP

III projects are not exclusively meant for poverty

eradication, and in case of irrigation projects, in-

tangible benefits will be more and the same are

not emphasized in the IEG study.

A gap ayacut of 10.07 lakhs is bridged during the

first three years of implementation of PIM. This

will definitely help the poor bring the livelihood

opportunities by way of increasing the number of

man working days.

3. A positive feature of irrigation devel-

opment since the 1997 reforms has been

rehabilitation of sections of the canal

network, so that water now reaches more

areas.

After formation of the WUAs and DCs in 1997,

massive O&M and MR works were taken in all

commands and farmers decided the works should

be taken up; they are actively involved in the ex-

ecution. However, there may be some defects in

certain cases; for example, WUA presidents did

not take other members into confidence and

benefited personally as contractors. This was cor-

rected, making the president’s election indirect,

which necessarily makes the decisions collec-

tive. WUAs are made continuous bodies. The

term of 5 years has resulted in complacency, so

the term of the president is 2 years to address this

problem.

The water rates could not be collected to the ex-

pected level due to continuous drought. It can be

observed that the water tax collections are around


1 1 3

100 Crores during 1998–99, 1999–2000, and

2000–01 as the season was good. Now O&M is

linked to the water tax collections, and the farm-

ers are being informed time and again that more

collections will bring sufficient funds for O&M.

It is not correct that the reforms have not broken

a vicious circle, as the study is limited to the SRSP

Project only, and there are no WUAs in SRBC;

thus, the study may not reflect the true picture of

PIM. The benefits of the PIM are:

• Sense of ownership

• Dynamic leadership

• Execution of works by WUA

• 10.07 lakh acres gap ayacut bridge

• Increased agricultural production.

As per the ORG study on the performance of the

WUAs, around 70 percent WUAs performance is

assessed above average in major and minor sec-

tors and 60 percent in medium sector [IEG note:

these figures are, in fact 50, 51, and 51 percent,

respectively; ORG 2005].

4. The farmers’ committee structure (WUAs,

DCs, and project committees) is intended

to ensure equitable allocation of water, but

the structure is incomplete and has lim-

ited capacity to do this.

During 1997 elections to the WUAs and DCs were

conducted, and they have taken over the charge

of O&M and water management. However, proj-

ect committees could not be formed, as it was felt

that the office bearers of the WUAs and DCs get

well experienced with the management of the ir-

rigation systems before they take up the re-

sponsibility of the project committee, which

requires better administrative and management

skills.

It is mentioned in the report that breaches of the

canal to take water are also common and WUAs

seem powerless to act. Indeed, they can act as ve-

hicles for elite capture of these resources. The

ORG study (2005) revealed that around 60 per-

cent of the farmers believe that the water supply

is more reliable and supply has increased.

Another study conducted by Dr. K.V. Raju (2001)

reports that in 50 percent of tail-end WUAs, farm-

ers felt the water has reached their fields with the

activities taken up by the WUAs. It is also reported

that there is considerable reduction in water

conflicts.

As these institutions are new, it cannot be ex-

pected that results will be at 100 percent. In the

first five years the WUAs and DCs actively involved

in the system maintenance. They have also par-

ticipated in water management. They have pri-

oritized and executed the work. The next step is

toward better water management. In this direc-

tion steps are being taken by the government

through training and capacity building.

5. Are WUAs an effective means to provide

sustainable O&M?

The O&M is linked to the water tax collections and

FOs. In the major and medium sector, FOs will get

50 percent, and in the minor sector 90 percent of

the water tax collections will go toward O&M. In

the previous year an amount of Rs 35 Crores was

released toward O&M and this year an amount of

Rs 60 Crores is earmarked for O&M. This will go

up to Rs 150 Crores in the coming years, and

O&M works can be taken up in all commands to

satisfactory level.

6. Bottom-up development needs support

from the top.

The government of Andhra Pradesh is keen to im-

plement PIM in the irrigation sector. As such,

elections to all the WUAs have been completed

and DCs are also constituted in certain projects;

in the remaining projects it will be completed by

the end of April 2007. As such, it is not correct that

there is a lack of support for WUAs, and govern-

ment is committed for stakeholders’ participa-

tion in management of irrigation systems.

7. WUAs have only limited means to re-

solve water-allocation disputes.

It is not correct to say that WUAs have only lim-

ited means to resolve water-allocation disputes per

the provisions of the Act. The farmers’ organiza-

tions have been given powers for settlement of dis-

putes (§§ 26, 27 of the Act).

1 1 4


V. Conclusions• Based on limited study, conclusions, if any are

drawn, will not be considered realistic.

• Visualizing the estimates either low or high is

probable in the sense that the features on the

ground cannot be foreseen accurately.

• Educating the farmers in a scientific way takes

much time. Processes are taking time.

• Farmers will be satisfied with the scientific

approaches only if they realize them on the

field.

• The success will depend on the availability of

water and the timing.

• The output/productivity largely depends on

many factors, like monsoons, pest control,

other management practices, and so forth.

• The farmers should get loans for their inputs, such

as for seeds, pesticides, fertilizers, and so forth.

• Only through sustained campaign and capac-

ity building activities can 100 percent water ac-

cess can be realized, leading to sustenance of

the systems.

• The reform process will continue in the irri-

gation sector, and the stakeholders will be

involved more in coming years for better man-

agement of the systems.


1 1 5

1 1 6


Attachment 1: Details of Ayacut Contemplated and Irrigated under SRBC in I to XVI Blocks by March 2007

Ayacut developed Ayacut to beBlock Contemplated with field developed with

Sl number number ayacut channels field channels

1 I 16,833.32 7,805 9,028.32

2 II 3,814.81 1,665 2,149.81

3 III 4,745.62 1,410 3,335.62

4 IV 8,869.92 2,090 6,779.92

5 V 4,324.91 1,200 3,124.91

6 VI 4,716.66 2,087 2,629.66

7 VII 13,170.44 7,000 6,170.44

8 VIII 6,333.41 1,500 4,833.41

9 IX 10,812.88 4,600 6,212.88

10 X 24,117.75 8,500 15,617.75

11 XI 4,520.17 3,944 576.17

12 XIA 4,073.11 3,832 241.11

13 XII 2,258.44 1,759 499.44

14 XIIA 5,805.33 4,885 920.33

15 XIII 13,248.68 9,885 3,363.68

16 XIV 9,021.23 8,464 557.23

17 XV 6,165.22 6,165.22

18 XVI 11,104.31 11,104.31

Total 153,936.21 87,895.53 66,040.68

Irrigation is the top priority area of Andhra

Pradesh. Jalayagnam is a unique venture of the

state. The strategy behind successful implemen-

tation of various projects that have been identi-

fied under Jalayagnam is detailed below.

For any scheme, provision of required funds and

their timely release is utmost importance. Budget

provided vis-à-vis expenditure incurred since

2004–05 speaks for itself.

Budget Expenditure

provided incurred

Year Rs in Crores Rs in Crores

2004–05 4,200 3,331

2005–06 6,300 6,569

2006–07 10,042 9,109

2007–08 13,014

All the project works are divided into convenient

packages under category I (Rs 150 Crores and

above), category II (Rs 50 Crores and above each

package), and open category and works are en-

trusted under an EPC turnkey system. The EPC

contractors are expected to do proper planning

and execute the works, as per the program. As

many as 245 packages have been awarded so far

and work is going on as programmed.

About 6.5 lakh acres of land is required to be

acquired for all the 35 major and 17 medium

projects. So far more than 2 lakhs acres of land

have been acquired on consent award method

after conducting Gram Sabhas. The district-level

committees are delegated with powers to de-

cide the rates up to 50 percent for prevailing

rates and the state-level committee with full pow-

ers. By this method land acquisition process is

speeded up.

Various departments of the government of India

are being pursued vigorously for speedy clear-

ances of projects.

Third-party quality control agencies are entrusted

with the job of overseeing the quality of con-

struction in all these projects.

An online monitoring system developed by the

Center for Good Governance is being used by all

the supervising officers right up to the secretaries

for review of progress of works.

Monthly review meetings by the Honorable Chief

Minister are held.

The irrigation sector requires a lot of improve-

ment. The irrigation efficiency in Andhra Pradesh

is only around 35 percent. It needs to be in-

creased to at least 50 percent by tight water man-

agement. Though PIM has been in vogue since

1997, the WUAs’ performance is not showing the

anticipated results in ensuring proper water sup-

plies to tail-end areas of the systems. This is a chal-

lenge that the irrigation sector is facing, and

every possibility is being explored to have effec-

tive water management.

As per the present estimate, an amount of Rs.

72,000 crores is required to complete the ongoing

35 major and 17 medium irrigation projects for cre-

ation of additional irrigation potential of 71.17

lakh acres. The expenditure incurred since June

2004 on irrigation projects is Rs 18,000 Crores.

Irrigation potential created in the last few years

is given below.

Year Ayacut in acres

237,886

220,491

692,950

Year Proposed ayacut in acres

1,003,630

1,582,239

1,099,499

Micro Irrigation: The government issued orders

to go for micro irrigation in all lift irrigation

schemes, including the Gaaleru Nagari Sujala Sra-

vanthi Scheme. The orders are issued primarily

with a view to improve the water use efficiency

while bringing additional area under irrigation in

certain cases.


1 1 7

Attachment 2: Innovative Measures to Speed Up the System to Ground a LargeNumber of Projects

1 1 8


IEG Response to Comments

Government comment IEG response

Irrigation is definitely a powerful force for poverty reduction.

To draw conclusions on the entire irrigation development projectsis inappropriate and not justifiable.

The methodology of assessing the impact, sample size, andselection of the sample chosen by IEG has significant differencesand therefore the findings are misleading. Conclusions made onSRBC based on the sample of SRSP are inappropriate.

It may not be out of context to mention that IEG has prejudged theissues and extrapolated/generalized the small sample surveyresults across the irrigation investments.

It is very much understandable that the two projects taken up areof different nature; that is, one is basically a rehabilitation of anexisting system (SRSP) and the other one is to complete a newirrigation system for bringing new areas under irrigation. Thus,special issues concerned with these two projects andcircumstances considered by the Bank for assisting are overlookedby the study and the report is made exclusive of these facts.

The government departments will be strengthened in due coursefor maintaining efficient management of the irrigation system, soas to redress weaknesses in the activities of WUAs.

The IEG has made drastic comments on the functioning ofgovernment departments. It is felt desirable not to make an issueof this sort of comments.

[IEG’s] conclusions are not based on project-specific data . . . [so]the conclusions of the study do not adequately represent reality.

Regarding the representation of the WUAs as mostly by the elitegroups, the study conducted by the IEG is only for a limited areaand for all practical purposes the study conducted by the ORG(2005) can be taken as reliable study, as the sample taken for thesame is 214 WUAs across the state and in all three sectors ofirrigation.

IEG fully agrees with this statement, the report providing newestimates of this impact.

The report is specifically an impact evaluation of Bank supportunder AP II and AP III. Where more general statements are made,they are supported by additional evidence, notably IEG’s reviewWater Management in Agriculture: Ten Years of World BankAssistance, 1994–2004 (IEG 2006b).

The government of Andhra Pradesh’s comment appears to assumethat IEG’s analysis and finding are based solely on the IEG survey.This survey was undertaken to be able to investigate the impact ofnew irrigation. But a wide range of sources, including governmentdata and the AP III “baseline,” are used in the study.

IEG does not agree with the suggestion that the issues wereprejudged. The issue of the sample was discussed in the previouscomment.

IEG’s report does distinguish between SRSP and SRBC, both interms of outputs and effectiveness. The report is clear that SRSP isrehabilitation and SRBC new construction, and that this is a majorreason SRSP has a higher rate of return than SRBC. However, IEG’srevisions after receiving the government’s comments have madethis distinction clearer.

In revising the report, IEG has in several places indicated that thegovernment of Andhra Pradesh has introduced measures toimprove the efficiency of construction.

IEG’s comments are substantiated by reference to other studieswith similar conclusions. As noted above, attempts by thegovernment of Andhra Pradesh to improve the administration ofirrigation are noted in IEG’s revised report.

Project-specific data are used for SRSP costs and benefits andSRBC costs. Because no water had been delivered to the bulk ofthe command area of SRBC at the time the study was undertaken,benefits are estimates similar to those in the Bank’s Implementa-tion Completion Report (the government’s comments of that reportdid not question these estimates).

IEG’s conclusions regarding WUA management are not based onits survey, but on a large survey of more than 200 WUAs carriedout by cess, and a number of detailed studies. The ORG study isnow referred to in the IEG report. The general conclusion is thatWUA committees are reasonably representative, but that thepresident is not, and often runs the WUA with little consultation.

(Table continues on next page)


1 1 9

IEG Response to Comments (continued)

Government comment IEG response

The water rates could not be collected to the expected levelbecause of continuous drought. It can be observed that the watertax collections are around 100 Crores during the periods 1998–99,1999–2000, and 2000–01, as the season was good.

It is not correct that the reforms have not broken a vicious circle, as the study is limited to the SRSP Project only, and there are noWUAs in SRBC.

It is not correct to say that WUAs have only limited means toresolve water-allocation disputes, per the provisions of the Act.The farmers’ organizations have been given powers for settlementof disputes (§§26, 27 of Act).

It is not correct that there is a lack of support for WUAs, andgovernment is committed to stakeholder participation inmanagement of irrigation systems.

In the revised report IEG gives the most recently available data onwater cess collection, which remains inadequate for O&M.

As noted above, cess collection remains lower than required, andproblems in water availability remain. The revised report gives thedate of WUAs being created in SRBC.

IEG recognizes these formal powers but maintains that they areinadequate to resolve water disputes, especially those betweencommunities.

The report has been revised to reflect the commitment of thegovernment to continuing the reform process.

1 2 1

Impact evaluation seeks to distinguish between

what happened with the intervention and what

would have happened in the absence of that

intervention—that is, with versus without. The

without is also called the counterfactual. Estab-

lishing a valid counterfactual is usually done using

a control group—that is, a group similar to the

treatment (beneficiary) group other than it does

not have the intervention.

One problem encountered in establishing the

counterfactual is that of possible selection bias,

whereby there is a systematic bias between treat-

ment and control communities. There are a num-

ber of ways around this problem. There are

experimental approaches that are either ran-

domized designs or natural experiments. Where

these are not possible or applicable, then a

quasi-experimental approach can be used, such

as regression discontinuity or propensity score

matching. Both these approaches are regression

based.

If selection is on observables, then regression-

based approaches can remove selection bias. If se-

lection is based on time-invariant unobservables,

then quasi-experimental approaches can also be

used if there are both baseline and endline data

(allowing double difference estimates). Only if

selection is on time-varying unobservables is it nec-

essary to use experimental approaches to avoid

selection bias.

IEG’s approach is to stress that impact evaluation

should be rigorous and relevant. That is, IEG’s

impact studies apply rigorous approaches but

also aim to deliver policy-relevant conclusions.

Doing this usually means using a theory-based ap-

proach that analyzes the whole results chain for

an intervention.

This impact study combines a variety of data

sources to examine impact. The survey used a

pipeline approach to identify the control com-

munities, and regression analysis for most in-

dicators. Other data sources allow estimates of

single, and in some cases, double-difference

estimates.

For further information on approaches to impact

evaluation, see the two IEG booklets Impact Eval-

uation: The Experience of the Independent Eval-

uation Group of the World Bank (IEG 2006a)

and Conducting Quality Impact Evaluation

under Budget, Time and Data Constraints (Bam-

berger 2006).

Recent IEG impact evaluations include: Agricul-

tural Extension: The Kenya Experience (2000);

Books, Buildings and Learning Outcomes: An Im-

pact Evaluation of World Bank Support to Basic

Education in Ghana (2004); and Maintaining

Momentum to 2015: An Impact Evaluation of Ex-

ternal Support to Maternal and Child Health

and Nutrition in Bangladesh (2005).

APPENDIX H: IMPACT EVALUATION

1 2 3

Chapter 11. Data are from FAOSTATS. For India these figures

are 2,070 and 2,470. About 2,000 calories is the mini-

mum daily requirement and is insufficient for those in-

volved in physical labor. Because these are average

figures, many people are below this level.

2. Gross cropped area “double counts” land used for

multicropping and additional cropping seasons; the lat-

ter in particular is often made possible by irrigation.

3. These findings are supported by several different

pieces of work, such as the econometric analysis with

district data of Datt and Ravallion (1998) and Fan and

Hazell (2001). A similar approach using state-level data

is Bhattarai and Narayanamoorthy (2003). Recent

project-level evidence comes from a study by the Inter-

national Water Management Institute, which collected

data from 26 irrigation systems in six Asian countries

(Hussain 2007). For a general overview on irrigation and

poverty, see Lipton, Litchfield, and Faurès (2003). For

a summary of recent Indian evidence see Malik (2005).

4. For example, average annual lending to irrigation

in the period 1994–98 was $610 million, compared

with $419 million from 1999 to 2004.

5. It has also been observed that the more rigorous

the ex post analysis, the lower the ERR (Berkoff 2002).

6. There are two aspects to excessive water con-

sumption. One is planting crops requiring large quan-

tities of water, notably paddy. The second is applying

more water than is strictly necessary.

7. The basic charge was raised from Rs 60/acre to Rs

200/acre, with some variations.

8. The legislation also put in place project commit-

tees at the scheme level, but these have not yet been

created.

9. WUAs are just one manifestation of Janmab-

hoomi. Many thousands of women’s self-help groups

were also formed. These are discussed in a separate re-

port on the Andhra Pradesh Rural Livelihoods Program

(World Bank 2004b).

10. The government of Andhra Pradesh states that

the delay occurred to allow a review of WUA perform-

ance and for necessary changes to be instituted.

11. AP II collected no baseline data. AP III collected

“baseline” survey data only in 2002, toward the end of

the project. This survey data is drawn on in the current

study.

12. The terminology here is the evaluation con-

vention of a project or treatment group (that is, those

receiving canal irrigation) and an untreated compari-

son group. The pipeline approach uses as the com-

parison group those who are scheduled to receive

the treatment but who have not yet done so, hence

overcoming the problem of selection bias caused by

drawing the comparison group from those who will not

participate. The bias arises because the determinants

of participation may be correlated with project

outcomes.

13. In its comments, the government of Andhra

Pradesh questions drawing conclusions for the whole

project based on a survey carried out in the tail end of

SRSP. However, as the text makes clear, IEG’s findings

are based on a wide variety of data sources.

14. A natural experiment means that a comparison

group arises out of a setting without intervention from

the evaluator.

15. In project villages, these village-level averages in-

clude households without direct access to the irrigation

system, but that may have benefited from the recharg-

ing of the groundwater through seepage from the canal,

tanks filled with canal water, and irrigated plots.

Chapter 21. Wade (1982) documents the channels for this

rent seeking in detail, and the amounts involved at

that time. It continues to be the case that engineers pre-

fer to work in construction rather than O&M.

2. See Vaidyanathan (1999) for a discussion of these

issues.

ENDNOTES

3. Even the ICR, having discussed the problems

that prevent water reaching the lower reaches and the

fact that these problems have not been addressed,

produced ERR estimates assuming the whole com-

mand will receive water.

4. The government of Andhra Pradesh piloted the

WUA approach in the SRSP command prior to AP III.

The expansion of WUAs across the command was, in

the words of the appraisal report, to be a “prototype”

for public investment schemes elsewhere in the state.

However, the “big bang” approach adopted for the

creation of WUAs meant that the support through AP

III did not serve as a prototype.

5. This figure is for distributaries up to km 234 only.

Under SRSP, a further 113,000 ha (funded by the gov-

ernment) have been added, from km 234 to km 284.

6. See also Vaidyanatan 1999 (pp. 59–61) on dis-

crepancies between different data sources regarding

irrigated area.

7. Revenue estimates of cropped area are the basis

for tax collection, so underreporting produces a rent to

be shared between the landowner and revenue official.

8. This problem is not unique to Andhra Pradesh or

India. IEG’s irrigation sector review reported figures for

four projects in East Asia for which the actual cropped

area was, on average, 67 percent of that expected at ap-

praisal (IEG 2002, table 3.4).

9. A partial caveat is in order here, because seep-

age from the canal raises groundwater levels and so fa-

cilitates groundwater irrigation (boreholes and tube-

wells, as well as collection in other water-harvesting

structures).

10. The SAR correctly states that the AP II investment

costs are sunk costs that are irrelevant to the decision

of whether to proceed with AP III. But in evaluating the

ex post return to irrigation investments, all costs should

be included. Assessing the overall investment based only

on the follow-on project would allow a sleight of hand

to produce a good rate of return. There is a parallel with

the multistate Subernarekha Irrigation Project from

the same time as AP II. This project financed initial

construction costs, with most of the irrigated area to

be established after the project. The appraisal report

discounted fears that the later investments may not take

place on the grounds that the initial investment costs

would by then be sunk costs, so the additional invest-

ment would have a very good rate of return! In the

event, cost overruns and delays meant that there was

no irrigation at all under the project and the Bank de-

cided not to finance a follow-on project.

11. In fact, the government’s own investment in

the scheme began in 1984, though these costs are not

shown in this analysis, so the delay is longer and the

return lower than shown here.

12. IEG (2002, p. 48–49) cites two other Bank

projects in India and one in Nepal where construction

delays contributed to a reduction in the ERR by 4

percent.

Chapter 31. One of the best known studies of community

management of natural resources is Wade’s (1988)

analysis of irrigation management in Andhra Pradesh.

2. In major schemes, 50 percent is meant to go to

the WUA, 20 percent to the distributary committee, and

20 percent to the project committee; in medium

schemes the WUA gets 60 percent, and the distributary

committee 30 percent; in minor schemes the WUA

gets the full 90 percent. The other 10 percent goes to

local government.

3. However, most of the discontent was about other

aspects of the reforms, such as the increase in the

price of subsidized rice and the reduction in power sub-

sidies. There were public protests on these issues and

the Communist Party condemned the Chief Minister as

being a “daily wage laborer of the World Bank” (Prasad

2004).

4. The preface to an early World Bank review of the

Andhra Pradesh experience stated, “The program [is]

part of an overall vision for alleviation of poverty being

applied in many sectors by Andhra Pradesh’s state lead-

ers.” More explicitly, the Web site of the International

Network on Participatory Irrigation Management says,

“PIM can enable the poor to have greater voice in de-

cisions. Conversely, if poor people are excluded and

their interests neglected, then irrigation development

may disrupt livelihoods and increase inequity. The

poor may suffer disproportionately from irrigation per-

formance problems such as water shortage in tail-end

areas, and so may also stand to benefit more from per-

formance improvements. Experience shows that par-

ticipatory irrigation management offers important

opportunities to empower the poor in good gover-

nance and to provide benefits for the poor.”

5. If a landowner does not engage in production him-

self, then only the tenants are eligible to be members.

1 2 4


6. Few farm households in the sample report no ir-

rigation of any sort on any of their plots. In 2005, there

were 40 such households (of 567 cultivating house-

holds), none of which was a WUA member.

7. There are various explanations for the lack of

political momentum. One is that the Chief Minister’s

attention became diluted with other concerns, notably

new schemes for the Godvari basin. Second, the Sec-

retary who had been instrumental in ushering in the

reforms left his post. Third, the new government was

suspicious of the new institutions created under the pre-

vious Chief Minister and so did not rush to implement

the already delayed elections.

8. Bodies such as WUAs have a potentially important

role to play when water is scarce, so scarcity might be

expected to make the associations more active as farm-

ers try to secure water. But the absence of water alto-

gether means there is nothing to be allocated.

9. The wealth index is constructed from data on own-

ership of consumer durables, jewelry, and housing

quality. It is used rather than income, as it is less affected

by endogeneity.

10. The population is divided into four groups:

scheduled castes, scheduled tribes, backward castes,

and other castes. The first two categories benefit from

reservation and other rights written into the consti-

tution. Backward castes were later identified as also

being disadvantaged groups requiring special sup-

port. The remainder, other castes, are the higher caste

groups.

11. It has been argued that this has not been the gen-

eral pattern in India in the past, as there has been a neg-

ative correlation between the “irrigation ratio” (the

proportion of a household’s land that is irrigated) and

total landholdings, though that negative correlation

has weakened in recent years (Vaidyanathan 1999, p.

85). But larger farmers have a larger absolute area of

irrigated land, resulting in a marked inequality: a World

Bank study reported that small and marginal farmers

accounted for 81 percent of agricultural households at

the national level but operate only 45 percent of the ir-

rigated land (World Bank 2003, p. 10).

12. The appraisal report for AP III envisaged that

WUAs would also branch into transportation, market-

ing, and agroprocessing. These things have not hap-

pened in any areas covered by this report.

13. This case illustrates one point (that the water flow

to the tanks is meant to be limited) but raises another

question (why the Irrigation Department has not op-

posed it elsewhere, raising the specter of political in-

fluence and rent-seeking).

14. This is the same figure as that found in another

study of WUAs in Andhra Pradesh (Alsop and others

2002).

15. The government of Andhra Pradesh disputes the

assertion that WUAs are subject to elite capture, citing

the ORG report (2005) as evidence. While it is true that

the WAU committees are more representative, the key

position of president is not. The lack of consultation

by the president is supported by the ORG report.

16. Other measures of socioeconomic status—

wealth, education, and an index of political status—are

correlated with landholding, so there are problems of

multicollinearity undermining their significance. If

land is dropped, then wealth and political status be-

come significant. A similar analysis is presented in the

paper of Alsop and others (2002), who asked WUA

members only. Large landowners, the non-poor, and

those who benefited from increased water supply

were more likely to respond that the WUA achieved ef-

fective management of the irrigation system and bet-

ter water distribution.

17. The same point is made in Mott MacDonald

(2005), drawing on experience from Nepal, the Kyrgyz

Republic, as well as Andhra Pradesh. They also em-

phasize that WUAs lack the technical expertise to man-

age the canal network.

18. This finding echoes that in Jairath’s (2000) re-

view of WUAs, written during the first term of office

when there was considerable optimism about the

Andhra Pradesh experience: “The tail-end areas of

the sample distributaries continue to be deprived of

their share of water. There was no effective way of pre-

vailing over the head villages even through the medi-

ation of distributary committees.” The World Bank

appraisal document for AP III stated that the forma-

tion of WUAs would create a lobby against water waste

in higher reaches, but this does not seem to have

happened.

19. In China, for example, where the benefits of ir-

rigation are more equitably distributed than they are

in India, Communist Party officials are active in the

water-allocation process. The same author says ex-

plicitly that “smooth functioning of the system requires

that there be some high authority to enforce rules and

resolve conflicts” (Vaidyanathan 1999, p. 27).

E N D N O T E S

1 2 5

20. However, government officials still had to sanc-

tion works proposed by the WUA and Reddy and Reddy

(2005) state that there was a lot of rent seeking asso-

ciated with this process, which is to be expected given

Wade’s analysis (1982).

21. This point is made in each of Mollinga, Do-

raiswamy, and Engbersen (2001), Alsop and others

(2002), and Jairath (2000).

22. According to Wade (1982), contractors play an

important role in the system of political and adminis-

trative corruption, being responsible for “paying off ”

so most of the rents raised do not have to pass through

the hands of senior politicians. The predominance of

contractors among WUA presidents supports suspi-

cions that the situation has not improved since Wade

wrote his paper.

23. There is one documented case of farmers in

Andhra Pradesh taking it upon themselves to form

their own project-level committee and claiming, albeit

unsuccessfully, the share of the project committee

(Mollinga, Doraiswamy, and Engbersen 2001).

24. On top of the water fee there is meant to be an

“own contribution” of 15 percent from WUAs to any

works carried out. These contributions are commonly

not collected, being estimated at just 5 percent (rather

than 15 percent) of expenditures.

Chapter 41. This is assuming average farm size of 1.3 ha in SRSP

(calculated from IEG survey, the SAR assumed 1.1 ha,

but the SRSP baseline gives a higher figure of 1.4 ha),

that the irrigation proportion is two-thirds, and that 80

percent of the reported ayacut (area served by the ir-

rigation scheme) actually receives irrigation water.

2. Cropping intensity can also increase with multi-

cropping, that is, growing more than one crop on the

same plot at the same time. However, multicrop-

ping is more common on rain-fed plots than irrigated

ones.

3. See appendix table D.8 for actual and expected

yields for major crops.

4. The World Bank’s ICR for AP III makes the same

observation.

5. “Double-difference” estimates of actual yield in-

creases were calculated using the survey data at both

plot and community levels. At the plot level, the first

difference is the average change in yields for plots that

grew paddy in both 2005 and 2006 but that used canal

or tank irrigation only in the latter year. This becomes

a double difference by subtracting from this yield in-

crease the yield increase in an “untreated” control area

(see appendix C). Two control groups were used. The

first control was a panel of plots that had not changed

their irrigation source between the two years. The sec-

ond control comprised plots that were not irrigated

during either year (no canal, tank, borehole, or tube-

well). In the latter case, the treatment group included

those with newly acquired canal or tank irrigation

who did not have irrigation in 2005. Although the

second double difference is a more satisfactory defi-

nition, the sample size becomes small, so the first

method has to be used for less common crops. This

analysis is complemented by a fixed effects regression

of the panel data.

6. The staff of the Agricultural Finance Corporation

assisted in the acquisition of these data.

7. The ICR yield increase estimates for paddy are

lower than this range.

8. This analysis was also performed controlling for

soil type, but the subsample sizes for the unirrigated

plots were too small to allow a meaningful comparison.

9. If the comparison group is well chosen, then the

treatment and comparison group have the same initial

conditions, so that the ex post single difference is equal

to the double difference.

10. It might be argued that farmers will take time to

realize the full yield potential. However, the yields of

newly irrigated farmers are not notably different from

those of already irrigated farmers, suggesting that there

is little scope for “catch-up.”

11. The shortfalls reported here are very similar to

those found in a review of four projects in East Asia, with

actual yields being 40, 47, 48, and 73 percent of those

expected at appraisal (IEG 2002, table 3.3).

12. The benefit stream is net income, that is, revenue

minus costs. The percentage change in net income is

the percentage change in revenue (–50 percent) times

the ratio of revenue to net income, which is taken as

1.5. Hence, the reduction in net income is 75 percent.

13. The appraisal ERRs are those given in the SAR,

which do not include the AP II sunk costs. According

to the SAR, taking these sunk costs into account reduces

the overall ERR by just over 5 percent and that for

SRSP below Lower Manair Dam by just over 3 percent.

Hence, the ERRs with a 50 percent reduction of yields,

and allowing for sunk costs, are 9 percent for SRSP

below Lower Manair Dam and 3 percent for the whole

project.

1 2 6


14. Although those with new access to canal or tank

irrigation appear more diversified, this may simply in-

dicate that farmers do not immediately adjust to the less-

diversified cropping pattern typical of those who have

access to irrigation water for a longer period. Moreover,

the double-difference estimates (comparing the change

in rice shares between 2005 and 2006 for households

that have newly acquired access to canal or tank irri-

gation) show an increase in rice share of 5–17 percent

for those who did not have irrigation previously and of

5–9 percent for those previously using other forms of

irrigation.

15. See Jairath 2000 for a discussion of the reasons

for rice preference.

16. In Andhra Pradesh the rice price for consumers

is guaranteed by the government. If the producer price

remains above this consumer price, then the effect of

the price reduction is a fiscal saving for government

rather than a benefit for consumers.

17. Vaidyanathan’s (1999) review of Indian irrigation

reports estimates in the range of 20– 30 percent.

18. In the plot-level regression for cropping inten-

sity, none of the irrigation source dummies is signifi-

cant once the number of sources is included in the

regression.

19. Three-quarters of plots with access to canal and

tank irrigation acquire the water by flooding, whereas

the majority of those with tubewell or borehole irri-

gation use an electric pump (appendix table D.14).

20. The IWMI study (Hussain 2007) concluded that

the multiplier in India was between 1.22 and 3, typically

around 2.21, which is higher than, but not inconsistent

with, Hazell and Haggblade’s (1990) finding that a Rs

100 increase in agricultural incomes increases non-

farm income by Rs 64, that is, a multiplier of 1.64. Bhat-

tarai, Barker, and Narayanamoorthy (2004) suggest an

even higher multiplier of 3–4.5.

21. A shadow wage rate of 0.5 has been used in this

report (in line with the appraisal and ICR calculations),

thus assuming that half the employment is additional

or that the alternative employment was only half as pro-

ductive.

22. Calculated for cultivating households only (see

appendix table D.15).

23. This explanation is supported by two findings.

First, in 2005 only 16 percent of cultivating households

hired out labor for agricultural purposes, but in 2006

nearly two-thirds (64 percent) did so. Second, the fixed

effects regression for domestic labor has a negative

coefficient on the canal dummy but a positive one on

the (canal x area) interactive variable, showing that in

larger households canal irrigation increases use of do-

mestic labor (see appendix C for results).

24. Using the constant 2005 price value of farm out-

put, labor productivity was Rs 101 per day’s labor in 2005

and 118 in 2006 (the calculation excludes farms with

failed crops).

25. The mean wage for women rose from Rs

24.95/day to Rs 27.50/day (and the median from Rs 25

to Rs 30). The mean for men rose from Rs 48.09 to Rs

50.72 (though the median remained unchanged at Rs

50); see appendix table D.24.

26. This increase in outside employment was al-

most entirely among cultivators: average outside em-

ployment rose by 100 days a year for cultivating families,

whereas there was no significant change for nonculti-

vating households (appendix table D.15), a finding

that mirrors the drop in the use of domestic labor.

27. For example, literacy in head reaches is higher

than in tail reaches, though the difference is not great

(48.0 versus 43.5 percent—calculated from census

data). The main difference is in the distance to Waran-

gal, which is on average twice as far for end-reach vil-

lages as for those at distributary heads (appendix table

D.26).

28. There is an important caveat that the analysis did

not include some benefits. First, it did not allow for the

more severe drought-related shocks on rain-fed plots.

Second, multiplier effects are not included. Third, the

welfare impact of lower food prices is not considered.

29. The benefit inflator is 13 percent (=1/0.88, not

12 percent) because the appropriate base is the share

used for canal irrigation. The assumption here is that

the benefit per unit of water is the same regardless of

end use, which is almost certainly not so, but as good

as any other assumption in the absence of data to

quantify the benefit from alternative uses.

30. The SRBC main canal is 141 km, which is to ir-

rigate 65,000 ha, which compares to the ayacut of up

to 323,000 ha covered by the 234 km of the Kakatiya

main canal in SRSP—that is, the irrigated area per km

of main canal in SRBC is one-third of that in SRBC.

31. The SAR for AP III did consider just one combi-

nation: the two construction-related risks of cost over-

run and delayed start.

32. This calculation subtracts from the ERR base

case return the decrease caused by allowing for sunk

costs (–5.2 percent), a 25 percent reduction in yields

E N D N O T E S

1 2 7

(–2.4 percent), a 20 percent increase in costs (–2.4 per-

cent), a two-year lag in benefits (–5.0 percent), and lim-

ited diversification (–5.4 percent).

Appendix A1. The head reach is the upper 25 percent in length

and the tail the bottom 25 percent. Reaches were iden-

tified by matching a village list against a list of WUAs,

giving villages and minor canals (subdistributary num-

bers) and using a system map to classify the minors by

reach. A small percentage of villages could not be iden-

tified (spellings of the same village can vary greatly by

the source consulted and even in different rounds of

the census) and so are listed as unclassified.

2. A village can be identified by its reach on each of

the main, distributary, and minor canals; for example,

a village with the position middle, head, tail, would be

a village in the middle reach of the main canal, at the

head of the distributary and toward the tail of the

minor. The reach referred to in the sample design is that

in the distributary canal. All sample villages fall within

approximately the same reach of the main canal, which

is the tail, though this was relatively well served by

water in 2005–06, given the good monsoon. The po-

sition of the minor also matters, as shown by the fact

that we found several instances in which villages at

the head of the minor captured all the water.

Appendix E1. It can be noted in passing that the very high cost

per irrigated hectare for SRBC requires substantial

yield increases to achieve an acceptable rate of return.

2. The poverty headcount actually falls by less under

scenario 2. This is because under scenario 2 the income

gain to those in the second quartile is less, so fewer cross

the poverty line. Although the income gain for those

in the first quartile is more under this scenario, it is in-

sufficient for them to reach the poverty line.

Appendix G1. Consolidated comments are prepared by IEG on

the basis of separate sets of comments submitted to IEG.

1 2 8


1 2 9

Alsop, Ruth, Disa Sjoblom, Ceema Namazie, and Pawan

Patil. 2002. “Community Level User Groups in Three

World Bank–Aided Projects: Do They Perform as Ex-

pected?” Social Development Paper No. 40, World

Bank, Washington, DC.

Bamberger, Michael. 2006. Conducting Quality Impact

Evaluations under Budget, Time and Data Con-

straints. ECD Series. Washington, DC: World Bank.

Belli. P., J. Anderson, H. Barnum, J. Dixon, and T. Jee-

Peng. 2002. Economic Analysis of Investment

Operations: Analytical Tools and Practical Appli-

cations. Washington, DC: World Bank Institute.

Berkoff, Jeremy. 2002. “Economic Evaluation: Why Is

It So Often Unsatisfactory? And Why Does This

Matter?” Paper presented at the International Con-

sulting Economists Association Meeting, London,

June 19.

Bhattarai, M., Randolph Barker, and A. Narayanamoor-

thy. 2004. “Who Benefits from Irrigation Develop-

ment in India? Implications of Irrigation Multipliers

for Cost Recovery and Irrigation Financing.” Paper

presented at the International Commission on Ir-

rigation and Drainage Asian Regional Workshop,

Taipei, November 10–12.

Bhattarai, M., and A. Narayanamoorthy. 2003. “Impact

of Irrigation on Rural Poverty in India: An Aggregate

Panel Data Analysis.” Water Policy 5: 443–58.

Datt, Gaurav, and Martin Ravallion. 1998. “Why Have

Some Indian States Performed Better than Others

at Reducing Rural Poverty?” Economica 65 (257):

17–38.

Fan, Shenggen, and Peter Hazell. 2001. “Returns to

Public Investments in the Less-Favored Areas of

India and China.” American Journal of Agricul-

tural Economics 83 (5): 1217–22.

Hazell, Peter, and Steven Haggblade. 1990. “Rural-

Urban Growth Linkages in India.” Agriculture and

Rural Development Department Working Paper,

World Bank, Washington, DC.

Hussain, Intizar. 2007. “Pro-Poor Intervention Strategies

in Irrigated Agriculture in Asia: Issues, Lessons, Op-

tions and Guidelines.” Irrigation and Drainage 56

(2–3): 119–26.

I&CADD (Irrigation & Command Area Development

Department). 2004. Third Andhra Pradesh Irriga-

tion Project: Project Completion Report. Hyder-

abad: Government of Andhra Pradesh.

IEG (Independent Evaluation Group). 2006a. Impact

Evaluation—The Experience of the Independent

Evaluation Group of the World Bank. Washington,

DC: World Bank.

———. 2006b. Water Management in Agriculture:

Ten Years of World Bank Assistance, 1994–2004.

IEG Study Series. Washington, DC: World Bank.

———. 2005. Maintaining Momentum to 2015: An Im-

pact Evaluation of External Support to Maternal

and Child Health and Nutrition in Bangladesh. IEG

Study Series. Washington, DC: World Bank.

———. 2004. Books, Buildings, and Learning Out-

comes: An Impact Evaluation of World Bank Sup-

port for Basic Education in Ghana. IEG Study

Series. Washington, DC: World Bank.

———. 2002. Bridging Troubled Waters: Assessing the

World Bank Water Resources Strategy. Washington,

DC: World Bank.

———. 2000. Agricultural Extension: The Kenya Ex-

perience. An Impact Evaluation. IEG Study Series.

Washington, DC: World Bank.

———. 1998. “Recent Experiences with Involuntary Re-

settlement: Overview.” Report No. 17538, World


Jairath, Jasveen. 2000. “Participatory Irrigation Man-

agement in Andhra Pradesh: Contradictions of a

Supply Side Approach.” Paper presented at South

Asia Regional Poverty Monitoring and Evaluation

Workshop, New Delhi.

Krishna, Anirudh, Mahesh Kapila, Sharad Pathak, Ma-

hendra Prowal, Kiranpal Singh, and Virpal Singh.

BIBLIOGRAPHY

2004. “Falling into Poverty in Villages of Andhra

Pradesh: Why Poverty Avoidance Policies Are

Needed.” Economic and Political Weekly 39 (20):

3249–56.

Lipton, Michael, Julie Litchfield, and Jean-Marc Faurès.

2003. “The Effects of Irrigation on Poverty.” Water

Policy 5: 413–27.

Malik, R.P.S. 2005. “Interventions in Development and

Management of Water and How They Impact on

Poverty.” Background Paper, World Bank, Washing-

ton, DC.

Mollinga, Peter, R. Doraiswamy, and Kim Engbersen.

2001. “Participatory Irrigation Management in

Andhra Pradesh, India.” International Journal of

Water 1 (3–4): 71–90.

Mott MacDonald. 2005. Equity, Irrigation and Poverty:

How to Distribute Water to the Poor. London: U.K.

Department for International Development.

Oblitas, Keith, J. Raymond Peter, Gautam Pingle, Halla

M. Qaddumi, and Jayantha Perera. 1999. “Transfer-

ring Irrigation Management to Farmers in Andhra

Pradesh, India.” Technical Paper No. 449, World


ORG Centre for Social Research. 2005. Performance

Evaluation of Farmers’ Organizations in Andhra

Pradesh. Hyderabad: I&CADD.

Plusquellec, Hervé. 2002. How Design, Management

and Policy Affect the Performance of Irrigation

Projects. Bangkok: Food and Agriculture Organiza-

tion of the United Nations.

Prasad, R.J. Rajendra. 2004. Emergence of Telugu

Desam. Hyderabad: Masterminds Academic Press.

Raju, K.V. 2001. “Participatory Irrigation Management

in India (Andhra Pradesh).” Paper for International

E-mail Conference on Irrigation Management Trans-

fer, June–October.

Reddy, V. Ratna, and P. Prudhvikar Reddy. 2005. “How

Participatory Is Participatory Irrigation Manage-

ment? A Study of WUAs in Andhra Pradesh.” CESS

Working Paper No. 65, Center for Economic and So-

cial Studies, Hyderabad.

Sivamohan, S.V.K., and C. Scott. 2005. “Coalition Build-

ing for Participatory Irrigation Management under

Changing Water Resource Trends: Reflections on

the Reforms in Andhra Pradesh.” In Asian Irrigation

in Transition: Responding to Challenges, ed. Ganesh

Shivakoti, Douglas Vermillion, Wai-Fung Lam, and Eli-

nor Ostrom. Thousand Oaks, CA: Sage Publications.

Thakkar, Himanshu. 1999. “Assessment on Irrigation in

India.” Contributing Paper of the World Commission

on Dams, London.

Upadhyay, B. 2004. “Gender Roles and Multiple Uses

of Water in North Gujarat.” Working Paper. Inter-

national Water Management Institute, Colombo.

Vaidyanathan, A. 1999. Water Resource Management:

Institutions and Irrigation Development in India.

New Delhi: Oxford University Press.

Wade, Robert. 1988. Village Republics. Cambridge,

U.K.: Cambridge University Press.

———. 1982. “The System of Administrative and Po-

litical Corruption: Canal Irrigation in South India.”

Journal of Development Studies 18 (3): 287–327.

White, Howard. 2006. Impact Evaluation: The Expe-

rience of the Independent Evaluation Group of

the World Bank. ECD Series. Washington, DC: World

Bank.

———. 2004. “Challenges in Evaluating Development

Effectiveness.” In Evaluating Development Effec-

tiveness, Vol. 7 of World Bank Series on Evaluation

and Development, eds. George Keith Pitman, Os-

valdo N. Feinstein, and Gregory K. Ingram, 33–54.

Edison, NJ: Transaction Publishers.

World Bank. 2005. India’s Water Economy: Bracing

for a Turbulent Future. Washington, DC: World

Bank.

———. 2004a. Water Resources Sector Strategy: Strate-

gic Directions for World Bank Engagement. Wash-

ington, DC: World Bank.

———. 2004b. “Women’s Self-Help Groups in Andhra

Pradesh—Participating Poverty Alleviation in Ac-

tion.” Working Paper No. 30831, World Bank, Wash-

ington, DC.

———. 2003. “The Incidence of Canal Irrigation

Subsidies in India: A Policy Note.” South Asia Rural

Development Unit Document, World Bank, Wash-

ington, DC.

———. 1997. “India—Water Resources Management

Sector Review: Urban Water Supply and Sanitation

Report.” Irrigation Sector Report, World Bank, Wash-

ington, DC.

1 3 0


WORKING FOR A WORLD FREE OF POVERTY

The World Bank Group consists of five institutions—the International Bank for Reconstruction and Development(IBRD), the International Finance Corporation (IFC), the International Development Association (IDA), theMultilateral Investment Guarantee Agency (MIGA), and the International Centre for the Settlement of InvestmentDisputes (ICSID). Its mission is to fight poverty for lasting results and to help people help themselves and theirenvironment by providing resources, sharing knowledge, building capacity, and forging partnerships in the publicand private sectors.

THE WORLD BANK GROUP

ENHANCING DEVELOPMENT EFFECTIVENESS THROUGH EXCELLENCE AND INDEPENDENCE IN EVALUATION

The Independent Evaluation Group (IEG) is an independent, three-part unit within the World Bank Group. IEG-World Bank is charged with evaluating the activities of the IBRD (the World Bank) and IDA, IEG-IFC focuses onassessment of IFC’s work toward private sector development, and IEG-MIGA evaluates the contributions of MIGAguarantee projects and services. IEG reports directly to the Bank’s Board of Directors through the Director-General,Evaluation.

The goals of evaluation are to learn from experience, to provide an objective basis for assessing the results of theBank Group’s work, and to provide accountability in the achievement of its objectives. It also improves Bank Groupwork by identifying and disseminating the lessons learned from experience and by framing recommendations drawnfrom evaluation findings.

THE INDEPENDENT EVALUATION GROUP

THE WORLD BANK SKU 175429 780821 375426

9 0 0 0 0

ISBN 978-0-8213-7542-6

an impact evaluation of india’s second and

Documents